RU2340114C1 - Coding device and method, decoding device and method and transmission system - Google Patents

Abstract

FIELD: information technology.

SUBSTANCE: in a wavelet transformation process, there is filtering with increment on lines, with a defined number of lines, as a result of which data coefficients for one line, corresponding the component of the lowest band, is generated in form of an increment. In the first time there is processing for line number 1 to line number 7, of coefficient C1 of the upper band component, coefficient C2 and coefficient C3, and coefficient Ca of the lower band component. Coefficient Cb and coefficient Cc, are generated at level 1 division, and at level 2 division. Coefficient C4 of the upper band component and coefficient C5 of the lower band component are additionally generated from coefficient Ca, coefficient Cb and coefficient Cc. These coefficients are reconfigured in order from the lower band to the upper band, and are transmitted to a synthesis filter. The synthesis filter filters the coefficients in the order in which they were transmitted. Images are generated and output.

EFFECT: coding with compression and decoding data images and output of decoded data images with little delay.

22 cl, 40 dwg

Description

FIELD OF THE INVENTION

The present invention relates to an encoding apparatus and method, a decoding apparatus and method and a transmission system, and in particular, relates to a data encoding apparatus and method, a data decoding apparatus and method, and a transmission system in which image data that is subjected to compression encoding and transmit, can be output as decoded images on the receiving side after a short time.

State of the art

Typically, an image compression encoding method is widely used, which is a JPEG (Joint Photographic Image Processing Experts Group) method that has been standardized by ISO (MOC, International Organization for Standardization). When using the JPEG method, image data is divided into blocks, and each of the divided areas is subjected to DCT (DCT, discrete cosine transform) processing to obtain DCT coefficients, and the obtained DCT coefficients are quantized and then entropy encoded, resulting in high quality and high compression ratio.

In addition, in recent years, studies have been made of a coding method in which an image is divided into multiple bands using filters called filter banks in which high-pass filters and low-pass filters are combined, and coding is performed for each of the divided banks. In each of these encoding methods, the encoding method, called wavelet transform encoding, in particular, is considered as a new technology, which is a likely candidate for replacing DCT encoding with image compression, because it does not have the problems that the DCT described above, which represents a block distortion at a high compression ratio.

For example, in JPEG 2000, the international standardization of which was completed in January 2001, a compression coding method is used in which the wavelet transform is combined with highly efficient entropy coding using a bit plane based on bit modeling and arithmetic coding. In this method, JPEG 2000 implements a significant improvement over conventional JPEG in terms of coding efficiency. Patent Document 1 describes a wavelet transform method in which coding efficiency has been further improved.

[Patent Document 1] Publication No. 9-130800 of an Examining Japanese Patent Application.

Such an international standard defines only the side of the decoder, and the side of the encoder can be developed arbitrarily.

SUMMARY OF THE INVENTION

Problems Solved by the Invention

It should be noted that in the JPEG method 2000 described above, for example, the usual necessary header information cannot be described in accordance with the algorithm until the coding of all pixels on the screen is completed. That is, such header information describes information necessary for decoding, such as the size of the compressed data obtained by encoding, and there is a need to add a marker at the end of the encoded data, but such information cannot be completed until all encoding is completed for one screen.

This problem is not unique to JPEG 2000, but also to JPEG, as well as to MPEG (Moving Image Expert Group), which is a method of compressing moving image data.

Accordingly, there is a problem arising from the use of conventional image data compression technologies, namely that the output of the encoded code stream, which is the result of compression encoding, must be performed after encrypting the image of the entire frame or field, in case of interlaced representation of the image .

Accordingly, there is a problem that cannot be solved, consisting in the fact that there is a time delay of one frame, or one field, in the case of an image with interlaced representation, between the transmission side and the receiving side of image data.

Accordingly, an object of the present invention is to provide an encoding apparatus and method, a decoding apparatus and method, and a data transmission system that allow compression encoding and decoding of image data and output of decoded image data with less delay.

Problem Solver

An encoding device in accordance with a first aspect of the present invention is an encoding device for encoding image data, including: filter means for hierarchically processing filtering image data and generating a plurality of subbands consisting of coefficient data divided by a frequency band; storage means for cumulatively storing coefficient data generated by the filter means; and coefficient reconfiguration means for reconfiguring coefficient data stored in the storage means for output in a predetermined order.

The filter tool may perform filtering processing in incremental rows starting from the upper side of the screen, image data towards the lower side.

The filter means may perform filtering processing of the image data in increments of blocks of rows that are image data consisting of a plurality of rows necessary to generate one row corresponding to coefficient data for at least a subband of a component of the lowest band.

The filter means may perform filtering processing both in the vertical direction and in the horizontal direction, in accordance with the image data.

At least one of a plurality of conclusions and a plurality of resolution levels in the filtering processing performed by the filter means may be determined in accordance with the target delay time.

The filter means may perform wavelet filtering processing in which the subband coefficient data of the lower band component obtained as a result of said filtering processing is further subjected to said filtering processing.

The filter tool may perform wavelet filtering processing using a lifting technique.

When the filter tool performs filtering processing of the level = X + 1 resolution using the rise technique, it can be performed on coefficient data calculated as a subband of the lower band component with filtering processing of the level = X resolution.

The storage means may further include: first buffer means for storing the subband coefficient data of the lower band component generated during processing of the wavelet filtering performed by the filtering means; and second buffer means for containing subband coefficient data of the upper band component generated during the wavelet filtering processing performed by the filter means.

The second buffer means may contain subband coefficient data for other band components except the lowest band until the subband coefficient data of the lowest band component is generated by the filter means.

The coefficient reconfiguration means may reconfigure the coefficient data so that the subbands are output in order from the lower band components to the upper band components.

The coefficient reconfiguration means may reconfigure the image data incrementally in blocks of rows, which are image data with the number of rows necessary to generate one row corresponding to the coefficient data for at least a subband of a component of the lowest band.

The encoding device may further include entropy encoding means for performing entropy encoding of coefficient data.

Entropy encoding means may sequentially perform entropy encoding of coefficient data reconfigured by coefficient reconfiguration means.

The coefficient reconfiguration means may reconfigure the coefficient data so as to output the subbands in order from the lower band components to the upper band components; wherein the entropy encoding means sequentially performs entropy encoding of the reconfigured coefficient data in an order from the components of the lower band to the components of the upper band as soon as the coefficient data is reconfigured by the coefficient reconfiguration means.

Entropy encoding means may perform entropy encoding of coefficient data generated by filtering means; wherein the storage means stores coefficient data obtained after entropy encoding using the entropy encoding means.

The storage means may store coefficient data generated as subbands of other components of the strip except the lowest band, and which were obtained after entropy coding by means of entropy coding, until the coefficient data for the subband of the component of the lowest band is entropy encoded with Entropy encoding tools.

The coefficient reconfiguration means may reconfigure the coefficient data stored by the storage means and obtained after entropy encoding with the entropy encoding means so that the subbands are output in order from the lower band components to the upper band components, and may output the reconfigured coefficient data in an order from the lower band components to components of the upper band after these coefficient data are reconfigured.

The entropy encoding means may perform group entropy encoding of multiple rows of coefficient data within a single subband.

The entropy encoding means can encode the rows of all subbands making up a block of rows that are a group of coefficient data corresponding to image data of a plurality of rows necessary to generate one row of coefficient data, at least for a subband of a component of the lowest band, with respect to the data stream coefficient, lined up in the one-dimensional direction, in order from the lower strip to the upper strip.

Entropy encoding means may include: quantization means for quantizing coefficient data generated by the filter means; and source encoding means for performing source encoding according to quantization result coefficients obtained from coefficient data quantized by quantization means.

The encoding device may further include: packetizing means for adding a predetermined header to the encoding result data obtained by the entropy encoding means performing entropy encoding, in order from the lower band components to the upper band components, for coefficient data of each block of rows that represent is a set of coefficient data corresponding to image data of a plurality of rows needed to generate one row of coefficient data and, at least, from the lowest sub-band component band, and packetizing the header and data body; and transmission means for transmitting a packet generated by the packetizing means; wherein the entropy encoding means, the packetizing means and the transmission means perform the corresponding processing simultaneously and in parallel; entropy encoding means performs entropy encoding of coefficient data in increments of blocks of rows; wherein the packetizer performs packetization of the encoding result data for each of the row blocks as soon as the encoding result data for each row block is generated as a result of entropy encoding performed by the entropy encoding means; and the transmission means transmits the received packet as soon as the encoding result data for each block of strings is packetized by the packaging means.

Identification information intended for identifying a block of lines on a screen, data body data length, and encoding information may be recorded in the header.

Entropy encoding means may include: quantization means for quantizing coefficient data generated by the filter means; and source encoding means for performing source encoding according to quantization result coefficients obtained for coefficient data quantized by the quantization means; wherein the encoding information includes a quantization step size for quantization performed by the quantization means.

The quantization step size information may include quantization step size information for each subband.

An encoding method in accordance with a first aspect of the present invention provides an encoding method for an encoding device for encoding image data, the method including: a filtering step for hierarchically performing filtering processing with respect to image data, and generating a plurality of subbands consisting of coefficient data distributed over the frequency band; a storage control step for cumulatively storing coefficient data generated as a result of processing in a filtering step in a storage module; and a coefficient reconfiguration step for reconfiguring the coefficient data stored in the storage unit, which is controlled by the processing performed in the storage control step for output in a predetermined order.

A decoding apparatus in accordance with a second aspect of the present invention is a decoding apparatus for decoding encoded image data in which image data has been encoded, including: storage means for storing coefficient data of a plurality of subbands consisting of coefficient data distributed over frequency band obtained by hierarchical processing of the first filtering of image data transmitted in incremental order cam; coefficient reconfiguration means for reconfiguring coefficient data stored by the storage means for output in a predetermined order; and filter means for performing second filtering processing of the coefficient data reconfigured with the reconfiguration means and outputted from the storage means and synthesizing the coefficient data from the plurality of subbands distributed over the frequency band to generate image data.

The coefficient reconfiguration means may reconfigure the coefficient data so that the subbands are output in order from the components of the lower band to the components of the upper band.

The coefficient reconfiguration means may reconfigure the coefficient data stored in the means of storage, incrementally in blocks of rows, which are a set of coefficient data corresponding to image data from a plurality of rows necessary to generate one row corresponding to the coefficient data, for at least subbands of the component of the lowest band.

The filter means may perform the second filtering processing in increments of rows starting from the upper side of the screen towards its lower side, thereby generating image data.

The filter tool performs the second filtering processing of the coefficient data in increments of blocks of rows that are a set of coefficient data corresponding to image data consisting of a plurality of rows necessary to generate one row corresponding to coefficient data for at least a subband of the lowest component stripes.

The filter means may perform a second filtering processing using a lifting technique.

The decoding device may further include entropy decoding means for performing entropy decoding of encoded data in row increments for each subband; wherein the storage means stores coefficient data obtained by entropy decoding performed by the entropy decoding means.

The entropy decoding means may decode encoded data, in which the rows of all subbands constituting the row block, which is a group of coefficient data corresponding to image data consisting of a plurality of rows necessary to generate one row of coefficient data of at least a subband of the component itself lower bands were encoded and lined up in one dimension.

Entropy decoding means may include: source decoding means for decoding a source for encoded data; and inverse quantization means for performing inverse quantization of coefficient data obtained by decoding the source using the source decoding means.

The decoding method of the second aspect of the present invention is a decoding method for a decoding device for decoding encoded image data in which image data has been encoded, the method including: a storage control step for storing a coefficient of a plurality of subbands in a data storage unit, distributed over the frequency band obtained as a result of hierarchical processing of the first filtering of image data transmitted from increment by row; a coefficient reconfiguration step for reconfiguring coefficient data stored in the storage unit under control performed by processing performed in the storage control step for output in a predetermined order; and a filtering step for performing second filtering processing with respect to coefficient data reconfigured using the processing performed in the reconfiguration step and outputting from the storage module and synthesizing coefficient data consisting of a plurality of subbands distributed over the frequency band to generate image data .

A transmission system in accordance with a third aspect of the present invention is a transmission system including: an encoding device for encoding image data, and a decoding device for decoding encoded data in which image data has been encoded; moreover, the transmission system is designed to transmit encoded data between the encoding device and the decoding device; in which the encoding device includes first filter means for hierarchically performing the first filtering processing of image data and generating a plurality of subbands consisting of coefficient data distributed over a frequency band, storage means for cumulatively storing coefficient data generated by the first filter means, and coefficient reconfiguration means for reconfiguring coefficient data stored by means with injury to output in a predetermined order; and in which the decoding device includes a second filter means for performing second filtering processing for coefficient data reconfigured by coefficient reconfiguration means and derived from the storage means that were transmitted from the encoding device through the data transmission and synthesis channel of the coefficient data consisting of a plurality subbands distributed over a frequency band for generating image data.

According to a first aspect of the present invention, filtering processing is performed hierarchically with respect to image data, wherein a plurality of subbands are generated consisting of coefficient data obtained by partitioning the frequency range, the generated coefficient data is stored and the coefficient data stored is reconfigured for output in a predetermined order .

In accordance with a second aspect of the present invention, coefficient data from a plurality of subbands obtained by partitioning a frequency range obtained as a result of hierarchical processing of the first filtering of image data transmitted in incremental rows is stored, the stored data is reconfigured to output it in a predetermined order, processing the second filtering is performed with respect to reconfigured and output coefficient data, and coefficient data of a plurality of subbands obtained as a result those frequency range partitions are synthesized to generate image data.

In accordance with a third aspect of the present invention, the first filtering processing is hierarchically performed on the encoding device for image data, a plurality of subbands are made up of coefficient data obtained by partitioning the frequency range, the generated coefficient data is stored and the coefficient data stored is reconfigured for output to a predetermined order; and in the decoding device, the second filtering processing is performed for coefficient data reconfigured in a predetermined order and transmitted from the encoding device through the transmission path, and coefficient data of a plurality of subbands obtained by partitioning the frequency range is synthesized to generate image data.

Benefits

In accordance with the present invention, filtering processing is performed hierarchically with respect to image data, and a plurality of subbands are generated, composed of coefficient data obtained by partitioning the frequency range, the generated coefficient data is stored and the coefficient data stored is reconfigured for output in a predetermined order, so so that during decoding the coefficient data can be processed in the order in which they arrive, and the delay time from encoding image data to decode encoded data and output the image data, preferably, can be reduced.

Brief Description of the Drawings

1 is a block diagram illustrating a configuration of an example image encoding apparatus to which the present invention is applied.

2 is a diagram describing a wavelet transform.

Figure 3 shows a diagram schematically describing a wavelet transform.

Figure 4 shows a diagram schematically describing a wavelet transform in the case of applying the method of raising the filter 5 × 3.

5 is a diagram schematically describing a wavelet transform in the case of applying the 5 × 3 filter raising technique.

6 is a diagram describing an example of performing filtering by rising using a 5 × 3 filter to a separation level = 2.

7 is a diagram describing a flow of wavelet transform and inverse wavelet transform in accordance with the present invention.

FIG. 8 is a flowchart for describing an example of encoding processing.

9 is a block diagram illustrating an example of an image decoding apparatus to which the present invention is applied.

10 is a flowchart for describing an example of decoding processing.

11 is a diagram describing the parallel operation of components of an image encoding device and an image decoding device to which the present invention is applied.

12 is a block diagram illustrating a configuration of an example image encoding apparatus to which the present invention is applied.

13 is a diagram describing a processing flow in the case of performing reconfiguration processing of wavelet coefficients on the side of the image encoding device.

Fig. 14 is a diagram describing a processing flow in the case of performing reconfiguration processing of wavelet coefficients on the side of the image decoding apparatus.

15 is a block diagram illustrating a configuration of an example image encoding apparatus in which the present invention is applied.

16 is a block diagram illustrating a configuration of an example image decoding apparatus in which the present invention is also applied.

17 is a diagram describing an example of how encoded data is exchanged.

On Fig shows a diagram illustrating an example configuration of a package.

19 is a block diagram illustrating an example configuration of an image encoding apparatus in which the present invention is applied.

On Fig shows a diagram describing the subbands.

21 is a diagram illustrating an example of quantized coefficients to be encoded.

22 is a block diagram illustrating an example configuration of an entropy coding unit.

23 is a flowchart of an encoding device.

24 is a flowchart for describing entropy encoding processing.

25 is a flowchart for describing encoding processing of a set w.

26 is a block diagram illustrating an example configuration of an image decoding apparatus.

27 is a block diagram illustrating an example configuration of an entropy decoding device.

28 is a block diagram illustrating an example configuration of a code separation module.

On Fig shows a block diagram illustrating an example configuration of a code separation module.

FIG. 30 is a flowchart for describing decoding processing.

FIG. 31 is a flowchart for describing entropy decoding processing.

32 is a flowchart for describing decoding processing of a set w.

33 is a block diagram illustrating another example configuration of an entropy coding unit.

34 is a diagram illustrating an example of quantized coefficients for encoding.

35 is a flowchart for describing encoding processing of a set w.

36 is a flowchart for describing decoding processing of a set w.

37 is a block diagram illustrating a configuration of an example of a digital triaxial system to which the present invention is applied.

38 is a block diagram illustrating an example configuration of a wireless transmission system to which the present invention is applied.

39 is a diagram illustrating an example configuration of a home game console to which the present invention is applied.

40 is a diagram illustrating an example configuration of an information processing system to which the present invention is applied.

List of Reference Items

1 image encoding device

10 wavelet transform module

11 intermediate calculation buffer module

12 coefficient reconfiguration buffer module

13 coefficient reconfiguration module

14 speed control module

15 entropy coding module

20 image decoding device

21 entropy decoding module

22 coefficient buffer module

23 inverse wavelet transform module

30 image encoding device

31 coding reconfiguration buffer module

32 coding reconfiguration module

41 image encoding device

42 image decoding device

43 coefficient reconfiguration buffer module

111 image encoding device

121 wavelet transform module

122 quantization module

123 entropy coding module

161 line definition module

162 efficiency coding module

163 maximum significant discharge calculation module

164 efficiency coding module

165 significant discharge highlighting module

166 efficiency coding module

167 character extraction module

168 efficiency coding module

169 code connection module

211 image decoding device

221 entropy decoding module

222 inverse quantization module

223 inverse wavelet transform module

251 code separation module

252 line definition module

253 generation module

254 efficiency decoding module

255 efficiency decoding module

256 efficiency decoding module

257 quantized coefficient synthesis module

258 switching module

271 control module

272 storage device

291 control module

401 buffer

500 transmission module

501 triax cable

502 camera control module

510 video encoding module

511 video decoding module

526 video decoding module

527 video coding module

600 transmission module

601 receiving device

602 video coding module

612 wireless module unit

621 wireless module unit

624 video decoding module

700 camcorder device

701 home game console

DETAILED DESCRIPTION OF THE INVENTION

The first embodiment in accordance with the present invention will be described with reference to the drawings. FIG. 1 shows one example of a configuration of an image encoding apparatus applicable to a first embodiment of the present invention. The image encoding device 1 comprises a wavelet transform module 10, an intermediate calculation buffer module 11, a coefficient reconfiguration buffer module 12, a coefficient reconfiguration module 13, a speed control module 14, and an entropy encoding module 15.

The input image data is temporarily stored in the intermediate calculation buffer module 11. The wavelet transform module 10 performs the wavelet transform of the image data that is accumulated in the intermediate calculation buffer module 11. That is, the wavelet transform module 10 reads image data from the intermediate calculation buffer 11 and performs filtering processing using the analysis filter to generate coefficient data with lower band components and upper band components and stores the generated coefficient data in the intermediate calculation buffer module 11. The wavelet transform module 10 has a horizontal analysis filter and a vertical analysis filter, and performs analysis filtering processing for a group of image data in both the horizontal direction of the screen and the vertical direction of the screen. The wavelet transform module 10 again reads the coefficient data of the lower band component, which is stored in the intermediate calculation buffer module 11, performs filtering processing on the coefficient data to be read using the analysis filter and additionally generated coefficient data using the upper band components and lower band components. The generated coefficient data is stored in the intermediate calculation buffer module 111.

The wavelet transform module 10 reads the coefficient data from the intermediate calculation buffer module 11 when this processing is repeated and the separation level reaches a predetermined level, and writes the read coefficient data to the coefficient reconfiguration buffer module 12.

The coefficient reconfiguration module 13 reads the coefficient data written into the coefficient reconfiguration buffer module 12 in a predetermined order and transmits to the entropy encoding module 15. Entropy encoding unit 15 encodes the transmitted coefficient data using an entropy encoding method, such as, for example, a Huffman encoding method or arithmetic encoding.

The entropy encoding unit 15 is controlled to work in conjunction with the rate control unit 14, in which the bit rate in the output after compression encoding is generally constant. That is, the speed control unit 14 transmits a control signal to the entropy encoding unit 15, in which the bit rate of the data bits encoded with compression by the entropy encoding unit 15 completes the encoding processing of the entropy encoding unit 15 at the point of reaching its target value or immediately before reaching its target value values based on information of encoding data from entropy encoding unit 15. Entropy encoding unit 15 outputs encoding data at a point at which encoding processing ends in accordance with a control signal transmitted from rate control unit 14.

The processing performed in the wavelet transform module 10 will be described in detail below. First, a general description of the wavelet transform will be given. Using the wavelet transform of the image data, which is generally shown in FIG. 2, processing is performed for dividing the image data into spatial frequencies from the upper bands and lower bands, recursively repeated with respect to data with a spatial frequency of the lower band obtained by the separation. Thus, by arranging the data with the spatial frequency of the lower band in a smaller area, compression coding can be performed more efficiently.

It should be noted that figure 2 shows an example for the case in which the separation processing in the component region of the lowest band of image data is repeated three times in the region L of the component of the lower band or region H of the component of the upper band, resulting in a level = 3 separation. In FIG. 2, “L” and “H” denote the components of the lower strip and the components of the upper strip, respectively, and the order “L” and “H” shows the strip as a result of horizontal separation of the front side and the strip as a result of vertical separation of the rear side. In addition, the numbers preceding “L” and “H” indicate the level of separation of their area.

In addition, as can be seen in the example shown in figure 2, the processing is performed in stages from the area in the lower right part of the screen in the direction of the area, the upper left part, into which the components of the lower band are transmitted. That is, in the example shown in FIG. 2, the four-part region in which the ZIN region in the lower right side of the screen, which has the components of the last lower band (includes most of the components of the upper band), has a region in the upper left parts of the screen, divided into four and further divided into four parts, and this area in the upper left part of the screen is further divided into four parts. The region in the farthest upper left corner is the 0LL region, which has the components of the lowest band.

The conversion and separation of the components of the lower band is performed repeatedly, since the image energy is concentrated in the components of the lower band. This can also be seen in the example of a subband formed as shown in FIG. 3B in accordance with the transition of the separation level from the separation level = 1 state, as shown in the example shown in FIG. 3A to the separation level = 3 state, as shown in the example shown in figv. For example, the separation level of the wavelet transform in FIG. 2 is 3, and therefore, 10 subbands are formed.

The wavelet transform module 10 typically performs the processing described above using a filter bank configured with a lower band filter and a high band filter. It should be noted that a digital filter usually has an impulse response, that is, a filter coefficient with a plurality of output lengths, and therefore, buffering of the coefficient data or input image data is required in advance only in the amount required to perform filter processing. In addition, similarly to the case of performing a wavelet transform using many steps, the wavelet transform coefficient generated in the previous step should be placed in the buffer only as many times as can be processed in this filter.

Next, as a specific example of the wavelet transform applicable in the first embodiment of the present invention, a method in which a 5 × 3 filter is used is described. A method in which a 5 × 3 filter is used is an excellent method in which wavelet transform can be performed with fewer filter taps, and is also used with the JPEG 2000 standard, as described with reference to a conventional technique.

The impulse response (Z-transform expression) of the 5 × 3 filter is made with a lower band filter H ₀ (z) and an upper band filter H ₁ (z), as shown in the following expressions (1) and (2). From the expressions (1) and (2), it can be determined that the filter H ₀ (z) of the lower band must have 5 taps, and it can be determined that H ₁ (z) of the upper band must have 3 taps.

In accordance with these expressions (1) and (2), the coefficients of the components of the lower band and the components of the upper band can be directly calculated. Now, using the lifting technique, you can reduce the number of calculations for filter processing. An overview of the analysis processing on the filter side using the wavelet transform, in the case of applying the lifting technique for a 5 × 3 filter, will be presented with reference to FIG. 4.

In Fig. 4, in each section of the uppermost stage, the middle stage section and the lower stage section, the columns of pixels of the input data, the output of the upper band component and the output of the lower band component are shown. At the uppermost stage, there is no need to be limited to the column of pixels of the input image, but the coefficient obtained from the previous filter processing can also be used. Here, the portion of the uppermost stage is a column of pixels of the input image in which a square mark indicates a pixel or line with an even number (starting at 0), and a round mark indicates a pixel or line with an odd number.

First, in the first step, a coefficient d _i ¹ of the upper band component is generated from the column of input pixels in accordance with the following expression (3).

Then, as a second step, a coefficient s _i ¹ of the lower band component is generated from the following expression (4) using an odd-numbered input image pixel.

Using the filter analysis side, the input image data is thus divided into lower band components and upper band components by filtering processing.

An overview of the synthesis side processing in the filter to perform the inverse wavelet transform, which restores the coefficient generated by the wavelet transform, will be described with reference to FIG. Figure 5 corresponds to figure 4 described above, it uses a 5 × 3 filter, and an example of the application of the lifting technique is presented. In Fig. 5, the portion of the uppermost stage represents an input coefficient generated using a wavelet transform, in which a round mark indicates the coefficient of the upper band and a square mark indicates the coefficient of the lower band.

First, at the first stage, in accordance with the following expression (5), a coefficient s _i ¹ with an even number (starting from 0) is generated from the input coefficients of the lower band component and the upper band component.

Then, in a second step, in accordance with the following expression (6), generating a coefficient d _i ⁰ of the odd-numbered coefficient s _i ⁰ with an even number, generated in the first step described above, and the ratio d _i ¹ input highband components.

Using the filter synthesis side, thus, in the filtering processing, the coefficient of the components of the lower band and the components of the upper band are synthesized, and the inverse wavelet transform is performed.

Next, a wavelet transform method in accordance with a first embodiment of the present invention will be described. FIG. 6 shows an example of filtering processing using a 5 × 3 filter lift, which was performed to separation level = 2, with reference to FIG. 4. 6, the portion shown on the left side of the circuit as an analysis filter is a filter for the wavelet transform module 10 on the side of the image decoding apparatus 1. In addition, the portion shown as a synthesis filter on the right side of the circuit is a filter for the inverse wavelet transform module on the side of the image decoding apparatus described below.

Using the description below with respect to a pixel in the upper left corner of the screen of a display device or the like as a leading pixel, it can be said that, for example, these pixels are scanned from the left end to the right end of the screen to configure one line, and line scanning is performed from the upper end of the screen towards the lower end, resulting in the configuration of one screen.

6, a column at the left end represents image data set so that it corresponds to a row of data of the original image aligned in the vertical direction. That is, filter processing with the wavelet transform module 10 is performed on pixels on a screen scanned vertically using a vertical filter. This shows the filter processing for the first column - the third column of level = 1 separation from the left end, and the fourth to sixth columns represent the processing of the filter level = 2 separation. The second column from the left end represents the upper band component output based on the image of the original image data on the left side, and the third column from the left end represents the lower band component output based on the original image data and the output upper band component. Filter processing, in which level = 2 separation, is performed as the output of filter processing for level = 1 separation, as shown in column 4 - column 6 from the left end.

In filter processing in which the separation level = 1, the upper band component coefficient data is calculated based on the original pixels of the image data as the first filter processing step, and the lower band component coefficient data is calculated based on the upper band component coefficient data calculated during the filter processing of the first stage and source pixels of image data. The filter processing for one example of separation level = 1 is shown in the first to third columns on the left side (filter analysis side) in FIG. 6. The calculated coefficient coefficient data of the upper band component are stored in the coefficient reconfiguration buffer module 12, as described with reference to FIG. In addition, the calculated coefficient data of the lower band component are stored in the intermediate calculation buffer module 11.

6, the coefficient reconfiguration buffer module 12 is represented as a portion surrounded by a dash-dot line, and the intermediate calculation buffer module 11 is shown as a portion surrounded by a dashed line.

Filter processing, in which the level = 2 separation, is performed based on the result of the level = 1 separation contained in the intermediate calculation buffer module 11. When processing a filter with separation level = 2, coefficient data calculated as coefficients of the lower band component during filter processing at separation level = 1 are used as coefficient data including lower band components and upper band components, and filter processing similar to filter processing for level = 1 separation. The coefficient data of the upper band component and the coefficient data of the lower band component, calculated by processing the filter with separation level = 2, are stored in the coefficient reconfiguration buffer module 12 described with reference to FIG. 1.

In the wavelet transform module 10, filter processing as described above is performed in each of the horizontal direction and the vertical direction on the screen. For example, first, filter processing with a separation level = 1 in the horizontal direction is performed, and the generated coefficient data with the upper band components and the lower band components is stored in the intermediate calculation buffer module 11. Then, filter processing is performed with the vertical level = 1 separation in relation to the coefficient data stored in the intermediate calculation buffer module 11. Using such processing in the horizontal and vertical directions, in which the separation level = 1, four regions are formed, which are the 'HH region and the HL region, each of which is formed using coefficient data, as a result of further separation of the upper band component into the upper band components and the components of the lower band, and the region LH and the region LL, each of which is formed by coefficient data, by further dividing the component of the lower band into the components of the upper band and the components of neither her band.

At a separation level = 2, filter processing is performed with respect to the coefficient data of the lower band components generated with the separation level = 1 in each of the horizontal direction and the vertical direction. That is, at the separation level = 2, the LL region formed by dividing at the division level = 1 is further divided into four regions, thereby additionally forming the HH region, the HL region, the LH region and the LL region within the LL region.

In the first embodiment, filter processing with a wavelet transform is performed step by step many times, dividing the processing in increments of several lines in the vertical direction of the screen. Using, for example, FIG. 6, in the initial processing used as processing starting from the first line on the screen, filter processing for seven lines is performed, and processing performed as the second processing starting thereafter from the eighth line performs filter processing in increments four lines each. The number of rows is based on the required number of rows to generate one row corresponding to the components of the lowest band, after splitting into two components of the upper band and the components of the lower band.

After that, the set of rows necessary to generate one row corresponding (coefficient data corresponding to one row of a subband of a component of the lowest band) to components of the lower band including other subbands is called a block of rows (or a section). Here, the row refers to one row that corresponds to pixel data or coefficient data generated within the image or field, or within each subband corresponding to the image data before the wavelet transform. That is, a row block (section) refers to a group of pixel data that corresponds to the number of rows needed to generate coefficient data corresponding to one row of a subband of a component of the lowest band after a wavelet transform of the original image data, before a wavelet transform, or a coefficient data group of each subband, obtained by wavelet transforming their pixel data groups.

In accordance with FIG. 6, coefficient C5 obtained from the filter processing of the separation level = 2 filter is calculated based on coefficient C _a stored in the intermediate calculation buffer module 11, and coefficient C4 is compared based on coefficient C _a , coefficient C _b, and coefficient C _s stored in the buffer 11 intermediate calculations. In addition, the coefficient C _{c is} calculated based on the coefficient C2 and the coefficient C3 stored in the coefficient reconfiguration buffer module 12, as well as on the basis of the pixel data in the fifth row. In addition, the coefficient C3 is calculated based on the pixel data of the fifth row - the seventh row. Thus, to obtain the coefficient C5 of the component of the lower band with the level = 2 separation, the pixel data in the first row - the seventh row is necessary.

And, on the contrary, when processing the filter for the second time and after that, it is possible to use coefficient data already calculated before the previous filter processing and stored in the coefficient reconfiguration buffer module 12, which reduces the required number of lines.

In other words, in accordance with FIG. 6, coefficient C9, which is the next coefficient following the coefficient C5, based on the coefficient C4 and the coefficient C8, and coefficient C _c stored in the intermediate calculation buffer module 11. The coefficient C4 has already been calculated using the filter processing described above, performed for the first time, and stored in the coefficient reconfiguration buffer module 12. Similarly, the coefficient _C has been calculated at the above-described filter processing performed for the first time, and saved in the module 11, midway calculation buffer. Accordingly, with such filter processing performed a second time, only filter processing is performed for a new calculation of coefficient C8. Such new filter processing is also performed using the eighth line — the eleventh line.

Thus, when filter processing is performed a second time, and after that, data already calculated before filter processing performed the previous time and stored in the intermediate calculation buffer module 11 and in the coefficient reconfiguration buffer module 12 can be used, which allows each time processing in increments of only four lines.

It should be noted that in the case when the number of lines on the screen does not match the encoded number of lines, the data lines of the original image are duplicated in a certain way so that they correspond to the encoded number of lines, after which filtering processing is performed.

Although a detailed description will be given below, in accordance with the present invention, the decoded image can be obtained with a slight delay, in the case of transmitting encoded data by performing phased filtering processing, to obtain only coefficient data corresponding to one row of a component of the lowest band divided into several times across the lines of the entire screen (in increments of blocks of lines).

To perform the wavelet transform, you need the first buffer used to perform the wavelet transform itself, and a second buffer designed to save the coefficient generated during processing up to the specified separation level. The first buffer corresponds to the intermediate calculation buffer module 11, and Fig. 6 is shown surrounded by a dash-dot line. The coefficient stored in the second buffer is used in the case of decoding, and thus represents the object of processing of entropy encoding in the next step.

The processing in the coefficient reconfiguration module 13 will be described below. As described above, the coefficient data calculated in the wavelet transform module 10 is stored in the coefficient reconfiguration buffer module 12, reconfigured by the coefficient reconfiguration module 13, and read and transmitted to the entropy encoding module 15.

As described above, when using the wavelet transform, coefficients are generated from the component side of the upper band to the component side of the lower band. In the example of FIG. 6, when executed for the first time, the coefficient C1 of the upper band component, coefficient C2, and coefficient C3 are sequentially generated when processing the filter with separation level = 1, from the pixel data of the original image. Filter processing with separation level = 2 is then performed for the lower band component coefficient data obtained by filter processing with separation level = 1, whereby a lower band component coefficient C4 and a coefficient C5 are successively generated. That is, the first time the coefficient data is generated in the order - coefficient C1, coefficient C2, coefficient C3, coefficient C4 and coefficient C5. The generation of coefficient data is always performed in this order (in order from the upper band to the lower band) based on the principle of wavelet transform.

Conversely, on the decoding side, for immediate decoding with low delay, it is necessary to generate and output an image from a lower band component. Therefore, reconfiguring the coefficient data generated on the encoding side from the component side of the lowest band to the component side of the upper band and transmitting them to the decoding side are preferred.

The following is a detailed description with reference to Fig.6. On the right side of Fig.6 shows the side of the synthesis filter that performs the inverse wavelet transform. Synthesis processing performed for the first time (inverse wavelet transform processing) including the first row of output image data on the decoding side is performed using coefficient C4, coefficient C5 and coefficient C1 of the lowest band component and generated by the filter processing performed in the first times on the coding side.

That is, when the synthesis processing is performed for the first time, the coefficient data is transmitted from the encoding side to the decoding side in the order of coefficient C5, coefficient C4 and coefficient C1, as a result of which, on the decoding side, processing is performed using coefficient C5 and coefficient C4 to generate coefficient Cf, using processing with a synthesis level = 2, which is a synthesis processing corresponding to a separation level = 2, and the coefficient Cf is stored in the buffer. The synthesis processing for the coefficient Cf and coefficient C1 is then performed with the synthesis level = 1 processing, which is a synthesis processing corresponding to the separation level = 1, whereby the first line is output.

Thus, when the synthesis processing is initially performed, the coefficient data generated on the encoding side in the order coefficient C1, coefficient C2, coefficient C3, coefficient C4 and coefficient C5, and stored in the coefficient reconfiguration buffer module 12, coefficient C5, coefficient C4 are reconfigured , coefficient C1, etc., and transmitted to the decoding side.

It should be noted that when using the synthesis filter side shown on the right side in FIG. 6, coefficients transmitted from the encoding side are denoted using the coefficient number on the encoding side inserted in parentheses, and they represent the line number of the synthesis filter outside the round brackets. For example, coefficient C1 (5) indicates that on the analysis filter side on the left side in FIG. 6, it represents coefficient C5, and on the synthesis filter side it represents the first row.

Synthesis processing on the decoding side using coefficient data generated by filter processing performed a second time, and then on the encoding side, can be performed using coefficient data transmitted from the synthesis in the case of synthesizing processing the previous time or on the coding side. In the example shown in FIG. 6, when synthesizing processing is performed a second time on the decoding side, which is performed using the coefficient C8 of the lower band component and coefficient C9 generated using filter processing performed on the second time, on the encoding side, it is further required decode the coefficient C2 and the coefficient C3 generated by the filter processing performed for the first time on the encoding side, and decode from the second to the fifth line.

That is, using the synthesizing processing performed a second time, the coefficient data is transmitted from the encoding side to the decoding side in the order coefficient C9, coefficient C8, coefficient C2, coefficient C3. On the decoding side, using the processing with the synthesis level = 2, the coefficient C _{g is} generated using the coefficient C8 and the coefficient C9, and the coefficient C4 transmitted from the encoding side in the synthesis processing performed for the first time. The coefficient C _{h is} generated using the facilitating coefficient C _g and the coefficient C4 described above, and the coefficient C _{f is} generated using the synthesizing process performed for the first time and stored in the buffer, and the coefficient Ch is stored in the buffer.

When processing with synthesis level = 1, synthesis processing is performed using coefficient C _g and coefficient C _h generated during processing with level = 2 synthesis and stored in buffer, coefficient C2 transmitted from the encoding side (represented as coefficient C6 (2) in synthesis filter), and coefficient C3 (represented as coefficient C7 (3) with a synthesis filter), and at the same time decode from the second line to the fifth line.

Thus, in the synthesis processing performed for the second time, coefficient data generated on the encoding side, such as coefficient C2, coefficient C3 (coefficient C4, coefficient C5), coefficient C6, coefficient C7, coefficient C8, coefficient C9, is reconfigured and transmitted to decoding side in order coefficient C9, coefficient C8, coefficient C2, coefficient C3, etc.

Thus, when using the third synthesis processing and after it, likewise, similarly, the coefficient data stored in the reconfiguration buffer module 12 is reconfigured in a certain way and transmitted to the decoding module, in which the lines are decoded in increments of four lines.

It should be noted that during the synthesis processing on the decoding side corresponding to the filter processing including the lines from the lower end of the screen on the encoding side, all coefficient data generated up to this point and stored in the buffer should be output, which, therefore, increases the number of output lines. According to the example shown in FIG. 6, eight lines were last displayed.

It should be noted that the coefficient reconfiguration processing by the coefficient reconfiguration module 13 sets the read addresses in the case of reading the coefficient data stored in the coefficient reconfiguration buffer module 12, for example, in a predetermined order.

The above processing will be described in detail with reference to Fig.7. Figure 7 shows an example of filter processing using wavelet transform up to level = 2 separation, which uses a 5 × 3 filter. With the wavelet transform module 10, as one example, shown at position A in FIG. 7, filter processing is performed for the first time for the first row — the seventh row of input image data in each of the horizontal and vertical directions (In - 1 at position A on Fig.7).

When filter processing is performed for the first time, at separation level = 1, coefficient data for three rows corresponding to coefficient C1, coefficient C2 and coefficient C3 is generated, and as one example, shown in position B in FIG. 7, each of them positioned in the HH region, the HL region, and the LH region formed with a separation level = 1 (WT-1 at position B in FIG. 7).

In addition, the LL region formed with the separation level = 1 is further divided into four parts when performing filter processing in horizontal and vertical directions with the separation level = 2. With coefficient C5 and coefficient C4 generated with division level = 2, one row is located in the LL region by coefficient C5 in the LL region with level = 1 division, and one row is located in each NI region, HL region and LH region, by C4 coefficient .

When using filter processing performed for the second time and then by the wavelet transform module 10, the filter processing is performed in four-line increments (IN - 2 ..., as shown in position A in Fig. 7), the coefficient data is incremented in two lines with level = 1 division, and coefficient data generates in increments one row with level = 2 divisions.

According to a second time example shown in FIG. 6, coefficient data corresponding to two rows for coefficient C6 and coefficient C7 is generated by filter processing at separation level = 1, and as one example, shown at position B in FIG. 7 , and positioned so that they follow the coefficient data that were generated by filter processing performed for the first time in the HH region, the HL region, and the LH region formed with a separation level = 1. In the LL region with the separation level = 1, the coefficient C9 corresponding to one row generated during filter processing with the division level = 2 is located in the LL region, and the coefficient C8 corresponding to one row is located in each of the HH region, the HL region, and the LH region .

In the case of decoding data subjected to wavelet transform, as shown in position B in Fig. 7, as one example, shown in position C in Fig. 7, the first line is output in the synthesis processing performed for the first time on the decoding side (Out- 1 at position C in Fig. 7), which corresponds to the filter processing performed for the first time using the first line, the seventh line on the encoding side. After that, four lines are simultaneously output on the decoding side (Out-2 at position C in Fig. 7), which corresponds to the filter processing a second time, until the last time on the encoding side. Eight lines are output on the decoding side, which corresponds to the filter processing, for the last time on the encoding side.

The coefficient data generated by the wavelet transform module 10 from the component side of the upper band to the component side of the lower band is sequentially stored in the module 12 of the coefficient reconfiguration buffer. In the coefficient reconfiguration module 13, when the coefficient data is accumulated in the coefficient reconfiguration buffer module 12, until it becomes possible to perform the coefficient reconfiguration described above, the coefficient data is reconfigured in the necessary order and read from the coefficient reconfiguration buffer module 12. The read coefficient data is sequentially transmitted to the entropy coding unit 15.

Entropy encoding unit 15 controls the encoding operation so that the bit rate of the output data becomes equal to the target bit rate based on the control signal transmitted from the rate control unit 14, and performs entropy encoding with respect to the transmitted coefficient data. The encoded data, after entropy encoding, is transmitted to the decoding side. As a coding method, a well-known technique such as Huffman coding or arithmetic coding can be considered. Of course, such a method should not be limited to these approaches, and any other coding method that allows inverse coding processing can be used.

It should be noted that if the entropy coding unit 15 first quantizes with respect to the coefficient data read from the coefficient reconfiguration unit 13 and performs encoding processing of the information source, such as Huffman coding, or arithmetic coding with respect to the obtained quantization coefficient, additional advantages can be expected in due to improved compression. As a method of such quantization, any method and, for example, a commonly used tool, that is, a means of dividing data for the W coefficient by the size A of the quantization step, as shown in the following expression (7), can be used.

As described with reference to FIG. 6 and FIG. 7, in a first embodiment in accordance with the present invention, the wavelet transform module 10 performs wavelet transform processing with sequentially incrementing a plurality of rows (incrementing row blocks) of image data. The encoded data encoded by the entropy encoding module 15 is output in sequential increments in the form of such blocks of strings. That is, in the case of processing, up to level = 2 separation, using a 5 × 3 filter to display one data screen, the output is received as one line for the first time, four lines for every second time and then to the last time, and eight lines are displayed for the last time.

It should be noted that in the case of replenishment for coefficient data after reconfiguration using module 13 for reconfiguring the entropy coding coefficient, for example, in the case of performing entropy coding for the row of the first coefficient C5 with filter processing performed for the first time, as shown, for example, in FIG. 6, there is no historical row, i.e. there is no row already generated by the coefficient data. Accordingly, in this case, only one line is subjected to entropy encoding. Conversely, in the case of encoding a string of coefficient C1, lines with coefficient C5 and coefficient C4 become historical rows. Such a plurality of strings approaching each other can be considered as strings that can be configured using similar data, thus performing entropy coding for a plurality of strings simultaneously is efficient.

In addition, as described above, with respect to the wavelet transform module 10, an example of performing filter processing with a wavelet transform using a 5 × 3 filter is described, but should not be limited to this example. For example, when using module 10 of the wavelet transform, you can use a filter with a large number of taps, such as, for example, a 9 × 7 filter. In this case, if the number of taps is greater, the number of lines accumulated by the filter also increases, as a result of which the delay time from inputting image data to outputting encoded data becomes large.

In addition, in the above description, the separation level of the wavelet transform has been described as separation level = 2 for the purpose of description, but should not be limited to this, and separation levels can be further increased. The stronger the separation level will be increased, the better and greater compression ratio can be realized. For example, in the general case, when using wavelet transform, filter processing is repeated up to level = 4 separation. It should be noted that as the separation level increases, the delay time also increases significantly.

Accordingly, in the case of applying the first embodiment of the present invention in an actual system, it is desirable to determine the number of filter taps or the separation level taking into account the delay time or image quality of the decoded image required in the system. The number of filter taps or the separation level need not have a fixed value, but they can also be selected accordingly.

Next, with reference to the flowchart of FIG. 8, a specific example of a processing stream for the entire encoding process in accordance with the image encoding apparatus 1, such as described above, will be described.

After the coding processing is started, in step S1, the wavelet transform module 10 sets the number A of the string block to be processed in the initial setting. In normal cases, number A is set to "1". After the installation is completed, in step S2, the wavelet transform module 10 obtains image data for the number of rows needed (i.e., one block of rows) to generate one row of the Ath row from the top of the subband of the lowest band, in step S3, the vertical analysis filter is processed for performing analysis filtering so that the image data is aligned in the vertical direction of the screen as its image data, and in step S4, the horizontal analysis filtering processing is performed to perform filtration analysis with respect to the image data arrayed in the horizontal direction of the screen.

In step S5, the wavelet transform module 10 determines whether analysis filtering processing has been performed to the last level, and if it is determined that the separation level has not reached the last level, the processing returns to step S3, in which the analysis filtering processing in step S3 and step S4 are repeated for the current split level.

In the case where it is determined in step S5 that the analysis filtering processing has been performed to the last level, the wavelet transform unit 10 transfers the processing to step S6.

In step S6, the coefficient reconfiguration module 13 reconfigures the coefficient of the row block A (the block of the A-th line forms the upper part of the image (fields, in the case of the interlaced representation method)) in the order from the lower band to the upper band. In step S7, the entropy encoding unit 15 subjects the coefficient to entropy coding in incremental increments. After the entropy encoding is completed in step S8, the entropy encoding unit 15 performs an external conversion of the encoded data of the row block A.

The wavelet transform module 10 sequentially increments the value for number A by “unit” in step S9, processing the next block of lines, and in step S10 determines whether or not raw input image lines for the given image exist (fields, in the case of the interlaced representation method) intended for processing. If it is determined that there are unprocessed input image lines, the processing returns to step S2, and then the processing is repeated for a new block of lines intended for processing.

As described above, the processing in steps S2-S10 is repeatedly performed to encode each block of strings. If it is determined in step S10 that there are no more raw image input lines, the wavelet transform module 10 ends the encoding processing for this image. A new encoding process begins for the next image.

In the case of the conventional wavelet transform method, the horizontal analysis filtering processing for the entire image is first performed (fields, in the case of the interlaced representation method), then the vertical analysis filtering processing for the entire image is performed. Similarly, the horizontal analysis filtering processing and the vertical analysis filtering processing are then sequentially performed for the entire lower band component obtained. As indicated above, the analysis filtering processing is repeated recursively until the separation level reaches the final level. In accordance with this, the results of each analysis filtering process should be contained in a buffer, but in this case, it is required that the buffer contains the results of filtering the entire image (field, in the case of interlaced representation) or the entire component of the lower band at the separation level at this point in time, which requires a large capacity storage device (the amount of data contained is very large).

In addition, in this case, if the wavelet transform for the entire image (field, in the case of the method of interlacing) does not end, it is not possible to reconfigure the coefficient or entropy coding in the next steps, which thus significantly increases the time delay.

Conversely, in the case of the module 10, the wavelet transform of the image encoding apparatus 1, the vertical analysis filtering processing and the horizontal analysis filtering processing are constantly performed in increments in the form of line blocks to the last level, as described above, so that in comparison with the usual method, the number the data that should be contained (placed in the buffer) at the same time (for one period of time) will be small, which significantly reduces the storage capacity allocated to the buffer. In addition, by performing analysis filtering processing to the last level, subsequent steps for processing coefficient reconfiguration or entropy coding processing can also be performed (i.e., coefficient reconfiguration or entropy coding can be performed incrementally in line blocks). Accordingly, the delay time can be significantly reduced compared to the conventional method.

FIG. 9 shows one example of a configuration of an image decoding apparatus corresponding to the image encoding apparatus 1 shown in FIG. 1. The encoded data output from the entropy encoding unit 15 (encoded data output in FIG. 1) of the image encoding apparatus 1 shown in FIG. 1 is transmitted to the entropy decoding unit 21 of the image decoding apparatus 20 in FIG. 9 (the encoded data input in FIG. 9), decoding is performed with respect to entropy coding, and coefficient data is obtained. The coefficient data is stored in module 22 of the coefficient buffer. In the inverse wavelet transform module 23, coefficient data stored in the coefficient buffer module 22 is used to perform synthesis filter processing using the synthesis filter, as described, for example, with reference to FIG. 5 and FIG. 6, and again the results of the synthesis filter processing are stored in module 22 of the coefficient buffer. The inverse wavelet transform module 23 repeats this processing in accordance with the separation level to obtain decoded image data (output image data).

Next, with reference to the flowchart shown in FIG. 10, a specific example of the processing flow of the entire decoding process performed by the image decoding apparatus 20, such as described above, will be described.

After the decoding process has been started in step S31, the entropy decoding unit 21 obtains encoded data, and in step S32, the encoded data is entropy decoded in row increments. In step S33, the coefficient buffer module 22 contains coefficients decoded and thus obtained. In step S34, the inverse wavelet transform module 23 determines whether a coefficient corresponding to one block of rows has been accumulated in the coefficient buffer module 22, and if it is determined that it has not been accumulated, the processing returns to step S31, after which the processing is performed and the inverse module 23 the wavelet of the conversion waits until a coefficient corresponding to one block of lines is accumulated by the module 22 of the coefficient buffer.

If it is determined in step S34 that the coefficient corresponding to one block of rows has been accumulated in the coefficient buffer module 22, the inverse wavelet transform module 23 transfers the processing to step S35 and reads the coefficient corresponding to one row block that is contained in the module 22 buffer coefficient.

The inverse wavelet transform module 23 exposes the read coefficient to the vertical synthesis filtering processing, in which the synthesis filtering processing is performed with respect to the coefficients aligned in the vertical direction of the screen, and in step S37, the horizontal synthesis filtering processing is performed, in which the synthesis filtering processing is performed with respect to the coefficients arranged in the horizontal direction of the screen, and in step S38 determines whether processing has been completed the synthesis filtering filter for level 1 (the level at which the separation level value is “one”), that is, it determines whether the inverse transformation was performed for the state before the wavelet transformation, and if it is determined that level 1 has not been reached, the processing returns to step S36, whereby the filtering processing in step S36 and step S37 is repeated.

In step S38, if it is determined that the inverse transform processing has ended to level 1, the inverse wavelet transform module 23 transfers the processing to step S39 and outputs image data obtained in the inverse transform processing.

In step S40, the entropy decoding unit 21 determines whether or not to complete the decoding processing, and if it is determined that the input of the encoded data is intermittent and that the decoding processing should not be completed, the processing returns to step S31, and the processing is then repeated. In addition, in step S40, in the case where the input of the encoded data ends, etc., so that the decoding processing ends, the entropy decoding unit 21 ends the decoding processing.

In the case of using the conventional inverse wavelet transform method, first, horizontal synthesis filtering is performed for all coefficients at the processed separation level in the horizontal direction of the screen, and then vertical synthesis filtering is performed in the vertical direction of the screen. That is, each time the synthesis filtering processing is performed, the result of the synthesis filtering processing needs to be contained in the buffer, but in this case, the synthesis filtering results for the separation level at this point in time and also all the coefficients at the next separation level are required in the buffer, which requires a large storage capacity (the amount of data to be contained is large).

In addition, in this case, the output of image data is not performed until all the inverse wavelet transforms are performed within the image (fields, in the case of interlaced representation), so the delay time from input to output is significantly increased.

Conversely, in the case of the inverse wavelet transform module 23 of such an image decoding apparatus 20, as described above, the vertical synthesis filtering processing and the horizontal synthesis filtering processing are performed continuously in increments of line blocks, up to level 1, resulting in, in comparison with in the usual way, the amount of data that needs to be placed in the buffer at the same time (for one period of time) will be small, which, thus, helps to reduce the amount of memory its device that you want to allocate for the buffer. In addition, when performing synthesis filtering processing (inverse wavelet transform processing), up to level 1, image data can be output sequentially until all image data within the image is obtained (in increments of string blocks), thus, compared with the usual way, the delay time can be significantly reduced.

The operation of the various elements of the image encoding apparatus 1 shown in FIG. 1 or the image decoding apparatus 20 shown in FIG. 9 (the encoding processing of FIG. 8 or the decoding processing of FIG. 10) is controlled in accordance with a predetermined program using , for example, not presented by the CPU (central processing unit). The program is stored in advance, for example, in a ROM not shown (read only memory). However, the device is not limited to this, and the entire device can operate by the interaction of synchronization signals or control signals between each element containing the image encoding device or image decoding device. In addition, an image encoding device or an image decoding device may be implemented using software running on a computer device.

Next, a second embodiment in accordance with the present invention will be described. In the second embodiment, an arrangement is used in which, in a system such as described above in the first embodiment, various elements of the image encoding apparatus 1 and the image decoding apparatus 20 operate simultaneously, thereby performing compression encoding processing and image decoding processing with less delay.

It should be noted that in accordance with the second embodiment, the image encoding apparatus 1 and the image decoding apparatus 20 described above in the first embodiment with reference to FIGS. 1 to 10, as well as the encoding method and decoding method, can be applied in the same way, therefore, their description will not be given here.

11 schematically shows simultaneously performed operations, for example, by various elements of an image encoding apparatus 1 and an image decoding apparatus 20, in accordance with a second embodiment of the present invention. 11 corresponds to the above FIG. 7. The WT-1 wavelet transform, performed for the first time, is performed (position B in FIG. 11) in the entropy encoding unit 15 with respect to the In-1 input image data (position in FIG. 11). As described with reference to FIG. 6, the wavelet transform WT-1, performed for the first time, starts at the point in time at which the first three lines arrive, and a coefficient C1 is generated. That is, from the moment of inputting the In - 1 image data to the beginning of the WT-1 wavelet transform, a delay corresponding to three lines is generated.

The generated coefficient data is stored in module 12 of the coefficient reconfiguration buffer. After that, the input image data is subjected to the wavelet transform, and when the processing performed for the first time ends, the processing proceeds to the WT-2 wavelet transform performed for the second time.

The In-2 image data is input for the second time WT-2 wavelet transform, and the WT-2 wavelet transform is performed simultaneously by reconfiguring the Ord-1 of three coefficients, which are coefficient C1, coefficient C4 and coefficient C5, using the module 13 reconfiguration of the coefficient (C in FIG. 11).

It should be noted that the delays from the end of the WT-1 wavelet transform to the start of Ord-1 reconfiguration can be delays corresponding to the configuration of a device or system, for example, delays associated with the transmission of a control signal representing an instruction to start the reconfiguration process to coefficient coefficient reconfiguration module 13 , the delay necessary to start processing performed by the coefficient reconfiguration module 13 with respect to the control signal, or the delay necessary to process the program s, and do not represent actual delay of the coding processing.

The coefficient data is read from the coefficient reconfiguration buffer 12, in the order in which the reconfiguration ends, passed to the entropy encoding unit 15 and subjected to the EC-1 entropy encoding (position D in FIG. 11). Entropy coding of EC-1 can begin without waiting for the end of the entire reconfiguration of the three coefficients; coefficient C1, coefficient C4 and coefficient C5. For example, at the time of the end of reconfiguration for one line from coefficient C5, which is output first, entropy coding may begin for coefficient C5. In this case, the delay from the start of the Ord-1 reconfiguration processing to the start of the EC-1 entropy encoding processing corresponds to one line.

The encoded data at which the EU-1 entropy encoding ends, performed by the entropy encoding module 15, is transmitted to the image decoding apparatus 20 through a certain transmission path (E in FIG. 11). As for the transmission path for transmitting encoded data, one can consider, for example, a data network such as the Internet. In this case, the encoded data is transmitted over IP (PI, Internet Protocol). However, the arrangement is not limited to this, and other encoded data transmission paths may be a data transmission interface, such as IEEE 1394 (IEEE, Institute of Electrical and Electronics Engineers), or a wireless data transmission channel introduced by IEEE 802.11 or the like.

After entering the image data, consisting of several lines corresponding to the processing performed for the first time, in the image encoding device 1, image data is sequentially input at the lower edge of the screen. In the image encoding apparatus 1, in accordance with the input of In-n image data (n is 2 or more), as described above, the WT-n wavelet transform, Ord-n reconfiguration, and EC-n entropy encoding are performed in four-line increments . Reconfiguration of Ord and EU entropy encoding for processing the last time in the image encoding apparatus 1 is performed for six lines. This processing is performed simultaneously in the image encoding device 1, as, for example, shown in position A in FIG. 11 — position D in FIG. 11.

The encoded data, which is encoded using the EC-1 entropy encoding in the image encoding apparatus 1, is transmitted via the data transmission path to the image decoding apparatus and transmitted to the entropy decoding module 21. Entropy decoding module 21 sequentially transmits the transmitted encoded data, which is encoded using the EC-1 entropy encoding to iEC-1, to entropy encoding and restores the coefficient data (position F in FIG. 11). The restored coefficient data is sequentially stored in the coefficient buffer module 22. After storing the coefficient data in the coefficient buffer module 22 in an amount that is only the amount that can be subjected to the inverse wavelet transform, the inverse wavelet transform module 23 reads the coefficient data from the coefficient buffer module 22 and performs the inverse wavelet transform iWT-1 using the read coefficient data ( position G in FIG. 11).

As described with reference to FIG. 6, the inverse wavelet transform module iWT-1 with the inverse wavelet transform module 23 may start operating at a point in time when coefficient C4 and coefficient C5 are stored in coefficient buffer module 22. Accordingly, the delay from the start of iEC-1 decoding introduced by the entropy decoding module 21 to the beginning of the inverse wavelet transform iWT-1 performed by the inverse wavelet transform module 23 corresponds to two lines.

After the inverse wavelet transform iWT-1 for these three lines corresponding to the end of the wavelet transform performed for the first time in the inverse wavelet transform module 23, Out-1 outputs image data generated by the inverse wavelet transform iWT-1 (position H in FIG. eleven). When outputting Out-1, the image data of the first line is output, as described with reference to FIG. 6 and FIG.

After entering the encoded coefficient data corresponding to three lines as a result of the processing performed for the first time in the image encoding apparatus 1, for the image decoding apparatus 20, coefficient data encoded in the entropy encoding EC-n (n is 2 or more) is sequentially input. The image decoding apparatus 20 performs iEC-n entropy decoding and iWT-n inverse wavelet transform with respect to the input coefficient data in four-line increments as described above, and Out-n outputting the reconstructed image data restored by the inverse wavelet is sequentially performed iWT-n conversions. The iEC entropy decoding and iWT inverse wavelet transform corresponding to the last time, image encoding devices execute for six lines and output Out eight lines. These processes are performed simultaneously in the image decoding device, as shown in the examples at position F in figure 11 - position H in figure 11.

As described above, by simultaneously performing various processing in the image encoding apparatus 1 and the image decoding apparatus 20, in order from the upper portion of the screen to its lower portion, image compression processing and image decoding processing may be performed with a slight delay.

The delay time from image input to image output, in the case of a wavelet transform, up to a separation level = 2, using a 5 × 3 filter, is calculated with reference to FIG. 11. The delay time from inputting the first row of image data into the image encoding apparatus 1 to outputting the first row of image data from the image decoding apparatus 20 is the total sum of the elements presented below. It should be noted that delays that differ in accordance with the system configuration, such as delays in the data transmission path or delays corresponding to the actual processing time of various parts of the device, are omitted here.

(1) Delay D WT from entering the first line to the end of the wavelet transform WT-1 corresponding to seven lines.

(2) Time D Ord, which corresponds to reconfiguring Ord-1 for just three lines.

(3) Time D EC, according to the entropy coding of EC-1, which corresponds to three lines.

(4) Time D iEC in accordance with the entropy decoding of iEC-1, which corresponds to three lines.

(5) Time D iWT in accordance with the inverse wavelet transform iWT-1, which corresponds to three lines.

Delays introduced by the various elements described above will be calculated with reference to FIG. 11. The delay D WT in point (1) corresponds to ten lines in time. Each D-Ord time in step (2), D EC time in step (3), D iEC time in step (4), and D iWT time in step (5) corresponds to the time of three lines. In addition, in the image encoding apparatus 1, entropy encoding of EC-1 can be started after the start of reconfiguration of Ord-1 of one line. Similarly, in the image decoding apparatus 20, the inverse wavelet transform iWT-1 can be started after two lines of entropy decoding iEC-1 have begun. In addition, iEC-1 entropy decoding can begin processing at the time of encoding, corresponding to the end of processing of one line in entropy encoding of EC-1.

According to this example, the time delay from inputting the first line of image data into the image encoding device to outputting the first line of image data from the decoding device corresponds to 10 + 1 + 1 + 2 + 3 = 17 lines.

A more specific example will be given to account for the time delay. In the case where the input image data is an interlaced video signal for HDTV (HDTV, high definition television), one frame is configured, for example, with a resolution of 1920 pixels per 1080 lines, and one field is 1920 pixels per 540 lines. Accordingly, if the frame rate is 30 Hz, one field consisting of 540 lines is input to the image encoding device 1 at a time of 16.67 ms (= 1 second / 60 fields).

Accordingly, the delay time for entering image data corresponding to seven lines is 0.216 ms (= 16.67 ms × 7/540 lines), which is an extremely short time, for example, compared with the time of renewal of one field. In addition, with regard to the total values of the delay D WT described above in paragraph (1), the time D Ord in paragraph (2), the time D EC in paragraph (3), the time D iEC in paragraph (4), and the time D iWT in paragraph (5), the number of processed lines is small, so the delay time is extremely reduced. If the elements performing various processing are arranged in the form of hardware, the processing time can be further reduced.

Next, a third embodiment of the present embodiment will be described. In the first and second embodiments described above, the coefficient data is reconfigured after performing the wavelet transform in the image encoding apparatus 1. In contrast, in a third embodiment according to the present invention, data reconfiguration is performed after entropy encoding. That is, in the image encoding apparatus, in this case, entropy encoding is performed for the coefficients generated after performing the wavelet transform for the input image data, and reconfiguration processing is performed for the data after entropy encoding. Thus, by performing reconfiguration of the coefficient data, after performing entropy coding, it is possible to reduce the amount of storage required for the coefficient reconfiguration buffer.

For example, in the case where the bit accuracy of the input image data is 8 bits, if the wavelet transform is performed to a multiple separation level, the bit accuracy of the generated coefficient data becomes approximately 12 bits, for example. If the coefficient reconfiguration processing is performed before the entropy encoding processing, the coefficient reconfiguration buffer module must save the specified number of lines corresponding to the coefficient data with a bit accuracy of 12 bits. As a result of compiling the coefficient data generated by the wavelet transform, which undergo reconfiguration processing after entropy encoding, the coefficient reconfiguration buffer can store data compressed as a result of entropy encoding, which requires only a small storage capacity.

12 shows an example of an image encoding apparatus according to a third embodiment of the present invention. The plot in Fig. 12, common with the above-described Fig. 1, has the same reference numerals and will not be described in detail.

The input image data is temporarily stored in the intermediate calculation module 11 of the image encoding device 30. The wavelet transform module 10 performs a predetermined wavelet transform with respect to image data stored in the intermediate calculation buffer module 11, as described with respect to the first embodiment. The coefficient data generated using the wavelet transform is passed to the entropy coding unit 15. The entropy encoding unit 15 works in conjunction with the data rate control unit 14 and is controlled so that the bit rate of the compression encoding output is approximately equal to a fixed value, and entropy encoding is processed with respect to the transmitted coefficient data. That is, entropy coding unit 15 encodes the obtained coefficients in the same order in which they were received, regardless of the order of the coefficients.

Coded data in which coefficient data generated by a wavelet transform is entropy encoded using entropy encoding unit 15 is temporarily stored in encoding reconfiguration buffer module 31. The encoding reconfiguration module 32 reconfigures and reads the encoded data from the encoding reconfiguration buffer module 31 until the encoded data to be reconfigured is stored in the encoding reconfiguration buffer module 31. As already described with respect to the first embodiment, the coefficient data generated by the wavelet transform module 10 is generated in order from the components of the upper band to the components of the lower band from the upper end of the screen toward the lower end. In order to output the image data on the low-delay decoding side, the encoding data stored in the encoding reconfiguration buffer module 31 is reconfigured in order from the components of the lower band to the components of the upper band of the coefficient data using the wavelet transform and read.

The encoded data read from the encoding reconfiguration buffer module 31 is transmitted to the transmission path, for example, as output encoded data.

It should be noted that the data encoded and output by the image encoding device 30 in accordance with the third embodiment can be decoded by the image decoding device 20 in accordance with the first embodiment described with reference to FIG. 9, similarly to the case described in the first embodiment fulfillment. That is, the encoded data input to the image decoding apparatus 20, for example, via a data transmission path, is decoded by entropy encoding in the entropy decoding unit 21, and the coefficient data is restored. The reconstructed coefficient data is sequentially stored in the coefficient buffer module 22. The inverse wavelet transform module 23 exposes the coefficient data stored in the coefficient buffer module 22 to the inverse wavelet transform and outputs image data.

Next, a fourth embodiment of the present invention will be described. In the first embodiment, the third embodiment described above, the reconfiguration of the coefficient data generated by the wavelet transform on the side of the image encoding device is performed, for example, as shown in FIG. 13. In contrast, in a fourth embodiment of the present invention, reconfiguration processing of coefficient data generated by the wavelet transform is performed on the side of the image decoding apparatus, as, for example, shown in FIG.

When processing the reconfiguration of the coefficient data generated by the wavelet transform, as described in the above third embodiment, a relatively large amount of storage in the coefficient reconfiguration buffer is required, and also high processing performance is required to perform the coefficient reconfiguration processing itself. In this case, also, if the processing performance on the side of the image encoding device is higher than a certain amount, even if the coefficient reconfiguration processing is performed on the side of the image encoding device, there are no problems, as described in the first to third embodiments.

Here are situations in which an image encoding device is installed in a device with relatively small computing capabilities, such as mobile terminals, such as cell phone or PDA (Pocket PC) terminals. For example, recently released products are widely used, in which the photo function is added to the cell phone terminal (called a cell phone terminal with a camera function). You can consider a situation in which image data, which is the image taken by a camera phone with a camera function, is subjected to compression encoding using wavelet transform and entropy encoding and transmitted via a wireless or cable data channel.

Such mobile terminals have limited processing capabilities of their CPUs and also have an upper limit on memory capacity. Therefore, the processing load using the coefficient reconfiguration described above is a problem that cannot be ignored.

Thus, as one example shown in FIG. 14, when constructing the reconfiguration processing on the side of the image decoding device, the load on the side of the image encoding device can be reduced, which allows installing the image encoding device in devices with relatively low processing capabilities, such as mobile terminal.

FIG. 15 shows an example configuration of an image encoding apparatus applicable to the fourth embodiment. It should be noted that in Fig. 15, the areas common with the one described above in Fig. 1 are denoted by the same reference numerals, and their detailed description is not given here.

The configuration of the image encoding device 41 shown in FIG. 15 is presented as a configuration in which the coefficient reconfiguration module 13 and the coefficient reconfiguration buffer module 12 are removed from the configuration of the image encoding device 1 shown in FIG. 1 above. In other words, in the fourth embodiment, the image encoding device 41 uses a configuration in which a combination of the wavelet transform module 10, the intermediate calculation buffer module 11, the entropy encoding module 15, and the bit rate control module 14 are used.

The input image data is temporarily accumulated in the intermediate calculation buffer module 11. The wavelet transform module 10 performs the wavelet transform of the image data incandescent in the intermediate calculation buffer module 11, and sequentially transmits the generated coefficient data in the order of the generated coefficient data to the entropy encoding module 15. That is, the generated coefficient data is transmitted to the entropy coding unit 15 in an order from the components of the upper band to the components of the lower band in accordance with the order of the wavelet transform. The entropy encoding unit 15 performs entropy encoding with respect to the transmitted coefficients, wherein the bit rate of the output data is controlled by the bit rate control unit 14. The coefficient data generated by the wavelet transform is output as encoded data that has been subjected to entropy encoding.

FIG. 16 shows one configuration example of an image decoding apparatus according to a fourth embodiment. It should be noted that in Fig.16 areas that are common with the above-described Fig.9 are denoted by the same reference numbers, and their detailed description is not given here.

The encoded data output from the entropy encoding unit 15 of the encoded image device 41 described in FIG. 15 is supplied to the entropy decoding unit 21 of the image decoding device of FIG. 16, is entropy encoded, and they become coefficient data. The coefficient data is stored in the coefficient reconfiguration buffer module 43 through the coefficient buffer module 22. After the coefficient data is accumulated in the coefficient reconfiguration buffer module 43 until the coefficient data is configured, the inverse wavelet transform module 23 reconfigures the coefficient data stored in the coefficient reconfiguration buffer module 43 in the order from the lower band components to the upper band components, and reads the coefficient data, then performs inverse wavelet transform processing using the coefficient data in a read order. In the case of using a 5 × 3 filter, its arrangement will be as shown in FIG. 14 described above.

That is, when processing from the start of one frame, for example, at the time of decoding of coefficient C1, coefficient C4 and coefficient C5, during entropy encoding stored in coefficient reconfiguration buffer module 43, the inverse wavelet transform module 23 reads the coefficient data from reconfiguration buffer module 43 coefficient and performs inverse wavelet transform processing. The data subjected to the inverse wavelet transform in the inverse wavelet transform module 23 is sequentially output as output image data.

It should be noted that in the case of the fourth embodiment, in the same way as described with respect to the second embodiment described above with reference to FIG. 11, simultaneously processing in various elements of the image encoding apparatus 41 and transmitting encoding data along the data transmission path and processing to various elements of the image decoding apparatus 42.

Next, a fifth embodiment of the present invention will be described. In the fifth embodiment, encoded data transmitted between the image encoding device and the image decoding device according to the first embodiment, the fourth embodiment, is arranged in packets.

On Fig shows a schematic diagram describing an example of how to perform the exchange of encoded data. In the case of the example shown in FIG. 17, similarly to the other embodiments described above, the image data is subjected to wavelet transform when inputted in the form of sequential increments in the form of blocks of lines corresponding to only a given number of lines (subband 51). If the specified level of separation of the wavelet transform of the coefficient line from the subband of the lowest band to the subband of the highest band is reached, reconfigured in the reverse order compared to the order in which they were generated, that is, in the order from the lower band to the upper band.

In subband 51 in FIG. 17, each of the sections divided into patterns of diagonal lines, vertical lines and wavy lines has different line blocks (as shown by arrows, the white space of subband 51 is also divided into consecutive increments of line blocks, and it is processed). The coefficients of the row blocks after reconfiguration are subjected to entropy encoding, as described above, while encoded data is generated.

Here, if the image encoding device transmits the encoded data in the same form, for example, difficulties may arise in the image decoding device associated with identifying the boundaries of various blocks of lines (or complex processing may be required). Thus, in the present embodiment, a layout is used in which the image encoding apparatus attaches the header to the encoded data in increments of, for example, blocks of lines, and transmits a packet formed from the header and encoded data.

In other words, after the image encoding device generates encoded data (encoder data) for the first block of lines (Lineblock-1), the encoded data is packetized and transmitted as a transmission packet 61 to the image decoding device, as shown in FIG. After the decoding device receives an image of a packet (received packet 71), its encoded data is decoded.

Similarly, after the image encoding device generates a second block of lines (Lineblock-2) of encoded data, the encoded data is packetized and transmitted as a transmission packet 62 to the image decoding device. After the decoding device receives an image of a packet (received packet 72), its encoded data is decoded. And again, similarly, after the image encoding device generates a third block of lines (Lineblock-3) of encoded data, the encoded data is packetized and transmitted as a transmission packet 63 to the image decoding device. After the decoding device receives a packet image (received packet 73), its encoded data is decoded.

The image encoding device and the image decoding device repeat the processing described above to the final Xth block of lines (Lineblock-X) (transmission packet 64, received packet 74). Thus, in the image decoding apparatus, a decoded image 81 is generated.

On Fig presents an example configuration of the header. As described above, the packet contains a header (Header) 91 and encoded data, Header 91 includes a description of the number (NUM) 93 of a row block and the length (LEN) 94 of the encoded data.

The image decoding apparatus can easily identify the boundaries of each block of lines by reading this information included in the header added to the received encoded data, thereby reducing the decoding processing load or processing time.

It should be noted that, as shown in FIG. 18, a size description (from Δ1 to Δn) of 92 quantization steps with increments in subbands configuring a block of rows can also be added. Thus, the image decoding apparatus can perform inverse quantization in increments in the subbands, which thus allows for additional detailed image quality control.

In addition, the image encoding device and the image decoding device can be arranged so that they simultaneously (in the form of trunk processing) perform the various processes of encoding, packetization, packet exchange and decoding described above, as described in the fourth embodiment, with size increments of block of lines.

Thus, the delay time until the output image is received in the image decoding apparatus can be significantly reduced. For example, FIG. 17 shows an example of operation using moving images presented alternately (60 fields / second). In this example, the time for one field is 1 second - 60 = approximately 16.7 ms, but due to the simultaneous execution of various processing, image output can be obtained with a delay time of approximately 5 ms.

Next, a sixth embodiment in accordance with the present invention will be described. In a sixth embodiment, a specific example of entropy encoding in an image encoding device and entropy decoding in an image decoding device in the various embodiments described above will be described. In the various embodiments described above, any entropy encoding method can be used, but when using the method presented in the present embodiment, the image encoding device can perform encoding with simpler calculations, which reduces the delay time, the use of computing resource and buffer storage capacity and etc.

It should be noted that, as described above, in various embodiments, encoding can be performed after quantization of the coefficient data in the case of entropy encoding, but the same can be said for the present embodiment, as a result, entropy encoding can be performed after quantization of the coefficient data, or data coefficients can be subjected to entropy coding without quantization. However, as will be described below, the method with performing quantization improves image quality, therefore, only entropy encoding with quantization performed will be described below. In other words, entropy encoding without quantization is not given in this description, but a description of entropy encoding with quantization performed can be used in it.

In addition, the reconfiguration of the coefficient may not be given below in the following description. In the various embodiments described above, a description is provided for the case of entropy encoding of reconfigured coefficient data, for the case of reconfiguring encoded data after entropy encoding, and for the case of reconfiguring coefficient data after entropy decoding. However, the reconfiguration used here is, in principle, a process of performing inverse wavelet transform processing at a high speed and is mainly not related to entropy encoding processing (and entropy decoding processing). In addition, even in the case of performing reconfiguration of the coefficient, its reconfiguration is performed within the blocks of rows, therefore, in principle, it does not affect the entropy coding, which will be described in the present embodiment, although its specific content will be described below. In other words, the entropy coding method according to the present example can be used similarly in the case of coding coefficient data whose order has been changed, or in the case of coding coefficient data before reconfiguration. Accordingly, a description of the reconfiguration of the coefficient will not be given to simplify the following description.

In other words, the entropy encoding according to the present embodiment will be described below for a case in which quantization processing is performed in the entropy encoding unit 15 of the image encoding apparatus 41 of the fourth embodiment shown in FIG. 15. It should be noted that for similar reasons, the image decoding device will also be described only in accordance with such an image encoding device, and no description will be given regarding cases of coefficient reconfiguration or without performing inverse quantization.

19 is a block diagram illustrating a configuration of an example image encoding apparatus using the present invention.

The image encoding device 111 has a wavelet transform module 121, a quantization module 122, and an entropy encoding module 123.

The wavelet transform module 121, for example, corresponds to the wavelet transform module 10 of FIG. 15 and performs similar processing. That is, the image (data) used as the signal of the component that has undergone the necessary DC level shift is introduced, for example, into the wavelet transform module 121.

The wavelet transform module 121 performs a wavelet transform of the input image and divides this image into multiple subbands. The wavelet transform module 121 transfers the wavelet subband coefficient obtained as a result of the wavelet transform to the quantization module 122.

The quantization module 122 quantizes the wavelet coefficients transmitted from the wavelet transform module 121 and transmits the obtained quantization coefficients as its result to the entropy encoding module 123.

Entropy encoding unit 123 performs entropy encoding with respect to quantization coefficients transmitted from quantization unit 122, and outputs the codes thus obtained as an encoded image (data). The images output from the entropy encoding unit 123 may, after performing bit rate control processing, for example, be packetized and recorded or transmitted to other devices (not shown) connected to the image encoding device 111.

That is, the quantization module 122 and the entropy encoding module 123 correspond, for example, to the entropy encoding module 15 and the bit rate control module 14 in FIG.

Next, with reference to FIG. 20 and FIG. 21, the entropy encoding module 123 of FIG. 19 will be described.

For example, as shown in FIG. 20, one subband is configured of six lines, from line L1 to line L6, and the position corresponding to a pixel on the line in the xy coordinate system is denoted as (x, y). Now, in each of the line diagrams, the x coordinate at the left end position is denoted by 0, and the y coordinate in the L1 row is denoted by 0.

The quantization coefficients in the expression in the bit plane and in each position (x, y) for the subbands are entered in raster scan order from line L1 to line L6, from quantization module 122 to entropy encoding module 123.

In other words, first, a quantization coefficient corresponding to the left end coordinate (0, 0) of the line L1 is input to the entropy coding unit 123. Next, the coefficient after quantization, corresponding to the coordinate located next to the right (1, 0) from the coordinate (0, 0), is entered into the entropy coding module 123, and the coefficients after quantization corresponding to the corresponding coordinates located next to the coordinates into which the coefficients were entered after quantization, they are sequentially introduced into entropy coding unit 123 until a coordinate is obtained at the right end of line L1. After entering all the coefficients that have been quantized in coordinates on the line L1, the coefficients after quantization corresponding to each coordinate on the line L2, from the coordinate of the left end of the line L2 (0, 1) sequentially to the coordinate on the right end, are entered into the entropy encoding module 123, and similarly, from row L3 to row L6, quantized coefficients corresponding to the coordinates on each row are input to entropy encoding unit 123.

For example, in FIG. 21, as shown in the upper left part of the diagram, after entering the twelve quantized coefficients into the entropy encoding unit 123, in order from the coordinate at the left end of the line L1 in FIG. 20, the entropy encoding unit 123 encodes the quantized incremental coefficients, equal to a given amount w (w = 4 in FIG. 21), which is predetermined.

Next, the quantized coefficients presented in the upper left part of Fig. 21, each of which is expressed in absolute values of its code, divided into binary digits (expression in the bit plane), and in the example shown in Fig. 21, the quantized coefficients "-0101" , "+0011", "-0110", "+0010", "+0011", "+0110", "0000", "-UN", "+1101", "- 0100", "+0111" and “-1010” on one line (line L1 in FIG. 20) is entered in this order into entropy encoding unit 123.

Each quantized coefficient consists of a code of a quantized coefficient expressed as "+" (positive) or "-" (negative) (hereinafter referred to as the Sign of quantized coefficients) and the absolute value of the quantized coefficient expressed in binary form. In FIG. 21, in bits representing the values of each digit of the absolute value of the quantized coefficients, the bit on the upper side in the figure represents the highest order bit (highest order bit). Accordingly, in the quantized coefficient “-0101”, its sign is “-”, and the absolute value expressed in binary form is “0101”, so the decimal representation of such a quantized coefficient is “-5”.

First, the entropy encoding unit 123 determines whether all quantized coefficients (absolute values) of one input string are equal to 0, and according to the determination results, outputs a code indicating whether all quantized coefficients of the string intended for encoding are equal to 0. If it is determined that all the quantized coefficients are 0, the entropy encoding unit 123 outputs 0 as a code indicating whether or not all quantized row coefficients are equal to 0 and ends the encoding of the quantized row coefficients currently being executed. In addition, in the case where it is determined that the value of all quantized coefficients is not 0 (not only for quantized coefficients equal to 0), entropy encoding module 123 outputs 1 as a code indicating that all or not quantized row coefficients are 0.

In the case where the twelve quantized coefficients shown in the upper left part of the drawing are introduced, the variable delay of the quantized coefficients of the input string is not only 0, then the entropy encoding module 123 outputs 1 in the form of a code, as shown in the upper right part of the drawing.

After code 1, indicating that not all quantized coefficients are equal to O, will be output as a code indicating whether or not all quantized row coefficients are 0, then entropy encoding module 123 encodes the first four (w) input quantized coefficients "-0101 "," +0011 "," -0110 "and" +0010 ".

Entropy coding unit 123 compares the maximum significant bits of four consecutive quantized coefficients entered at this point in time (the value of variable B in FIG. 21) with the maximum significant bits of four (w) quantized, encoded (entered) the previous time, determines whether the maximum significant digit is changed, and displays a code indicating the maximum significant bits of the quantized coefficients.

Now, the maximum significant bits represent the maximum significant bits of the quantized coefficient having the largest values of the four (w) quantized coefficients that must be encoded together. In other words, the maximum significant digit indicates, with the quantized coefficient having the largest absolute value, of the four quantized coefficients, in which digit is the 1 with the highest order. Accordingly, the maximum significant bit of the four quantized coefficients "-0101", "+0011", "-0110" and "+0010", which must be encoded together, for example, is equal to "3", which is a bit, which contains 1 of the highest order for the quantized coefficient "-0110", which has the largest absolute value.

In addition, the code denoting the maximum significant bits of the quantized coefficients consists of a code identifying whether the maximum significant bit has changed, a code indicating whether the maximum significant bit has increased or decreased, and a code indicating the magnitude of the change in the maximum significant bit, in the case where the maximum the significant digit has not changed, a code indicating whether the maximum significant digit has increased or decreased, and a code indicating the magnitude of the change in the maximum significant digit is not output.

In the case when the maximum significant digit has changed, according to the results of comparing the maximum significant bits, entropy encoding module 123 outputs a code 1 indicating that the maximum significant digit has changed, and in the case where the maximum significant digit has not changed, displays a code 0 indicating that the maximum significant digit has not changed.

In addition, with regard to determining whether or not the maximum significant digit has changed, in the case when the four quantized coefficients are introduced for the first time, that is, in the case when the first quantized coefficients of the subband intended for encoding are introduced (for example, in the case when four quantized coefficients are entered on the left side, in the order of the line L1 of Fig. 20), while there are no quantized subband coefficients that were previously encoded, therefore, the maximum significant bit of the previously encoded four (w) kva The adjusted coefficients are set equal to 0.

Accordingly, entropy encoding module 123 compares the maximum significant bit of the four quantized coefficients “-0101”, “+0011”, “-0110” and “+0010” entered this time, which is 3, with the maximum significant bit four quantized coefficients encoded the previous time, which is 0, and outputs code 1 because the maximum significant digit has changed.

In addition, entropy encoding unit 123 outputs a code indicating whether the maximum significant digit has increased or decreased after code 1, indicating that the maximum significant digit has changed. Here, entropy encoding unit 123 outputs 0 when the maximum significant bit has increased, and outputs 1 when the maximum significant bit has decreased.

The previous maximum significant digit was 0, and the current maximum significant digit was 3, therefore, as shown in the upper right part of the drawing, entropy encoding module 123 outputs a code 0 indicating that the maximum significant digit has increased.

In addition, after outputting a code indicating whether the maximum significant bit has increased or decreased, the entropy coding unit 123 outputs a code indicating how much the maximum significant bit has increased or decreased, using, for example, a code indicating the magnitude of the change in the maximum significant bit. In particular, when the magnitude of the change in the maximum significant digit (that is, when the magnitude of the increase or the magnitude of the decrease) is n, the entropy encoding unit 123 outputs the (n-1) code Os and outputs the code 1 after these Os.

In the case of encoding the first four quantized coefficients shown in FIG. 3, the magnitude of the change in the maximum significant digit is 3 (= 3-0), as a result of which the entropy encoding module 123 outputs two (= 3-1) Os, and additionally outputs 1 to quality of code.

Then, entropy coding unit 123 outputs a code for maximum significant digits indicating the absolute value of each of the four (w) quantized coefficients that must be encoded at this time. That is, entropy coding unit 123 outputs, for each quantized coefficient, a code indicating the value of each digit for the absolute value of the quantized coefficient, from the largest digit of the significant digits indicated by the maximum significant digit, in order to the smallest digit.

The quantized coefficients to be encoded at this time are "-0101", "+0011", "-0110" and "+0010", so entropy encoding module 123 first outputs a code for the maximum significant digits indicating the absolute value of the quantized coefficient "-0101" entered first. Here, the maximum significant digit is 3, therefore, the entropy coding module 123 outputs the value “1” of the largest digit from the significant digits indicated by the maximum significant digit of the quantized coefficient “-0101” (that is, the third digit), the value “0” of the digit is one less than the largest discharge (second discharge) and the value "1" for the smallest other discharge. Thus, the code "101" is derived for significant digits indicating the absolute value of the quantized coefficient "-0101".

In the same way, entropy encoding module 123 outputs in this order the codes “011”, “110” and “010” for significant digits indicating the absolute values of the quantized coefficients “+0011”, “-0110” and “+0010”. Accordingly, "101011110010" is output as a code for the maximum significant digits indicating the absolute values of each of "-0101", "+0011", "-0110" and "+0010". Thus, the entropy code encoding unit 123 outputs a code with a length corresponding to the maximum significant bits of the four quantized coefficients, which should be encoded as a code indicating the absolute values of the quantized coefficients.

Finally, entropy coding unit 123 outputs a code denoting the sign of each of the four (w) quantized coefficients in which the absolute value is not 0. Now, in the case where the sign of the quantized coefficient is “+” (positive), entropy coding unit 123 outputs code 0, and in the case where the sign of the quantized coefficient is “-” (negative), it outputs code 1.

The quantized coefficients to be encoded this time are "-0101", "+0011", "-0110" and "+0010", and the signs of these quantized coefficients are in order negative, positive, negative, positive, as shown in the upper right part of the drawing, the entropy coding unit 123 outputs “1010” as a code indicating the sign of each of the quantized coefficients.

After encoding the four quantized coefficients introduced for the first time, the entropy encoding module 123 again encodes the next four consecutive quantized coefficients of “+0011”, “+0110”, “0000” and “-0011”.

In the same way as in the case of encoding the quantized coefficients introduced first (the previous time), the entropy encoding module 123 first compares the maximum significant bit of the four (w) quantized coefficients newly introduced this time with the maximum significant bits of four (w) quantized coefficients encoded the previous time.

The maximum significant digit of the four (w) quantized coefficients introduced this time, "+0011", "+0110", "0000" and "-0011", is "3", which is the discharge in which 1 of the highest order is located , for the quantized coefficient "+0110" having the largest absolute value, and it has the same value as the maximum significant bits of the quantized coefficients encoded the previous time, therefore, the entropy encoding module 123 outputs a code 0 indicating that the maximum significant bit is not has changed.

Further, entropy encoding module 123 outputs a code “011110000011” in which the code for the maximum significant digits “011”, “110”, “000” and “011” denotes the absolute values of each of the four (w) quantized coefficients “+0011", "+0110", "0000" and "-0011", which should be encoded this time, arranged in that order.

After outputting a code indicating the absolute values of the quantized coefficients, the entropy coding unit 123 outputs a code indicating the sign of each of the four (w) quantized coefficients, the absolute value of which is not 0.

The quantized coefficients to be encoded this time are “+0011”, “+0110”, “0000” and “-0011”, and the third quantized coefficient “0000” has an absolute value of zero, resulting in an entropy module 123 encoding outputs the code "001", indicating the signs of these quantized coefficients that are not equal to 0, "+0011", "+0110" and "-0011" (positive, positive, negative).

After encoding the four quantized coefficients "+0011", "+0110", "0000" and "-0011", the entropy encoding module 123 then encodes the following four consecutive quantized coefficients of "+1101", "-0100", "+0111" and " 0101 ".

Entropy coding unit 123 first compares the maximum significant bit of the four (w) quantized coefficients newly introduced this time with the maximum significant bits of the four quantized coefficients encoded the previous time.

The maximum significant digit of the four (w) quantized coefficients introduced this time, "+1101", "-0100", "+0111" and "-0101" is "4", which is the discharge in which 1 of the highest order is located for the quantized coefficient “+1101” having the largest absolute value, and it differs from the maximum significant bits “3” of the quantized coefficients encoded the previous time, as a result of which the entropy coding unit 123 outputs a code 1 indicating that the maximum significant digit has changed.

In addition, the previous maximum significant digit was 3, and the maximum significant digit this time was 4, as a result of which the entropy encoding module 123 outputs a code 0 indicating that the maximum significant digit has increased, as shown on the right side of the drawing.

In addition, entropy coding unit 123 outputs a code indicating how much the maximum significant digit has increased. In this case, the magnitude of the change in the maximum significant digit is 1 (= 4-3), as a result of which the entropy coding module outputs 0 (= 1-1), zeros, and additionally outputs 1 (that is, outputs code 1).

Next, entropy encoding module 123 outputs the code “11010100011010”, in which the code for the maximum significant digits is “1101”, “0100”, “0111” and “1010”, indicating the absolute values of each of the four (w) quantized coefficients “+1101", "-0100", "+0111" and "-1010", which should be encoded this time, was arranged in the indicated order.

After outputting a code indicating the absolute values of the quantized coefficients, the entropy coding unit 123 outputs a code indicating the sign of each of the four (w) quantized coefficients for which the quantized coefficient is not 0.

The quantized coefficients to be encoded this time are "+1101", "-0100", "+0111" and "-010", and the signs of these quantized coefficients are in this order positive, negative, positive, negative, as shown in the lower right part of the drawing, the entropy coding unit 123 outputs a code “0101” indicating the sign of each of the quantized coefficients.

Entropy encoding unit 123 thus encodes a sequential predetermined number (w) of input quantized coefficients. Thus, after a code indicating whether or not all quantized coefficients in a row intended for encoding were 0, and a code indicating that not all quantized row coefficients are 0, are derived from entropy encoding module 123, then a code is output denoting the maximum significant digits for w quantized coefficients, a code denoting the absolute values (expression in the bit plane) of w quantized coefficients, and a code denoting the signs of the quantized coefficients.

Regarding the code denoting the maximum significant bits of these w quantized coefficients, the code denoting the absolute values of w quantized coefficients and the code denoting the signs of the quantized coefficients, the code denoting the maximum significant bits of the next w quantized coefficients, the code denoting the absolute values of w quantized coefficients , and a code denoting the signs of the quantized coefficients is repeatedly output until all the quantized coefficients in this row are encoded.

The following is a description of the encoding of the quantized coefficients in raster scan order, but the order in which the quantized coefficients encode does not have to be a raster scan order. For example, in the case of encoding the quantized subband coefficients shown in FIG. 20, an arrangement can be made in which the quantized coefficients are encoded at (0, 0), (0, 1), (0, 2) and (0, 3) (i.e., in the positions from the left end in the drawing of each of the line L1 to the line L4), and the following quantized coefficients are encoded in positions (1, 0), (1, 1), (1, 2) and (1, 3) and so on, with the quantized coefficients in four positions vertically aligned in the drawing, used as w quantized coefficients and encoded in the specified time d w simultaneously.

Entropy encoding module 123, shown in FIG. 1, which performs processing such as described above, is implemented as shown in more detail in FIG.

Entropy encoding module 123 includes a string determination module 161, an efficiency coding module (VLC, variable length code) 162, a maximum significant discharge calculation module 163, an efficiency encoding module 164, a significant discharge allocation module 165, an efficiency encoding module 166, an allocation module 167 sign, efficiency encoding module 168 and code connection module 169.

The quantized coefficients output from the quantization module 122 (FIG. 19) are transmitted (input) to the string determination module 161, the maximum significant digit calculation module 163, the significant digit allocation module 165 and the sign extraction module 167.

Row determination module 161 determines whether all quantized coefficients of one row intended for encoding entered from quantization module 122 are equal to 0, and provides information indicating the results of this determination to efficiency encoding module 162.

Based on the information indicating the results of the determination made using the row determination module 161, the efficiency encoding module 162 outputs a code indicating whether all quantized row coefficients intended for encoding are equal to 0 to the code connection module 169.

The maximum significant digit calculation module 163 calculates the maximum significant bits for w continuous quantized coefficients input from the quantization module 122 and provides information indicating the results of its calculations to the efficiency encoding module 164 and the significant digit extraction module 165.

Based on the information indicating the calculation results from the maximum significant discharge calculation module 163, the efficiency coding module 164 transmits a code indicating the maximum significant discharge for w quantized coefficients to the code connection module 169.

Based on the information indicating the calculation results from the maximum significant discharge calculation unit 163, the significant discharge allocation unit 165 extracts a significant discharge for w quantized coefficients transmitted from the quantization module 122 and transmits (data) about the selected significant bits of the quantized coefficients to the encoding module 166 Efficiency and character extraction module 167.

The efficiency encoding module 166 encodes the absolute values of these quantized coefficients based on the significant bits of the quantized coefficients from the significant bit allocation module 165 and transmits the absolute values of the quantized coefficients resulting in the code connecting module 169.

The sign extraction module 167 extracts the signs of the quantized coefficients transmitted from the quantization module 122, based on the significant bits of the quantized coefficients, from the significant bit allocation module 165 and transmits (data) the extracted characters to the efficiency encoding module 168.

The efficiency encoding module 168 encodes (data) extracted characters transmitted from the character extraction module 167, and transmits a code indicating the signs of the quantized coefficients thus obtained to the code connection module 169.

Code connection module 169 connects a code indicating whether all quantized row coefficients are equal to 0, a code indicating maximum significant bits, a code indicating the absolute value of the quantized coefficients, and a code indicating the signs of the quantized coefficients transmitted from the efficiency encoding module 162, encoding module 164 Efficiency, Efficiency coding module 166 and Efficiency coding module 168, respectively, and outputs as an encoded image (data).

Next, the encoding processing performed by the image encoding apparatus 111 (Fig. 19) will be described with reference to the flowchart shown in Fig. 23. This encoding processing starts after the image (data) intended for encoding is input into the wavelet transform module 121.

In step S111, the wavelet transform module 121 subjects the input image to the wavelet transform, splits the input image into multiple subbands, and transmits the wavelet coefficients of each subband to quantization module 122.

In step S112, the quantization module 122 quantizes the wavelet coefficients transmitted from the wavelet transform module 121, and transmits the resulting quantized coefficients to the entropy encoding module 123. Thus, the quantized coefficients of each of the positions in the subband in the bit plane expression, which has been described with reference to, for example, in FIG. 21, are input to entropy encoding unit 123.

In step S113, the entropy encoding unit 123 performs entropy encoding processing and ends the encoding processing. Although the entropy encoding processing will be described in detail below, in the entropy encoding processing as described with reference to FIG. 21, the entropy encoding module 123 simultaneously encodes a sequential predetermined number (w) of quantized coefficients transmitted from the quantization module 122 and outputs it as encoded image (data), a code indicating whether all quantized coefficients are equal to 0 in the line that is intended for encoding, a code indicating the maximum significant bits w of the quantized coefficient in the code indicating the absolute values of the quantized coefficients, and code indicating the signs of the quantized coefficients.

Thus, the image encoding apparatus 111 encodes and outputs an input image.

Next, with reference to FIG. 24, entropy encoding processing corresponding to the processing performed in step S113 of FIG. 23 will be described.

In step S112 shown in FIG. 23, the quantized coefficients output from the quantization module 122 are passed (input) to the row determination module 161, the maximum significant digit calculation module 163, the significant digit allocation module 165, the sign extraction module 167, the entropy module 123 encoding (Fig.22).

In step S141, the row determination module 161 receives a variable y denoting a subband string, which should now be encoded as y = 0, and stores this value.

For example, in the case of encoding the quantized subband coefficients shown in FIG. 20, row determination module 161 takes a variable y denoting rows (row L1 to row L6) as y = 0. It should be noted that the string y, denoted by the variable y, denotes the string in which the coordinate y in each position (x, y) on the row of the subband is equal to y. Accordingly, in the case where the variable y, which the string determination module 161 stores, is, for example, y = 0, the string denoted by this variable becomes the string L1, in which the coordinate y in each position of the string is 0.

In step S142, the maximum significant digit calculation module 163 receives the Binit variable denoting the maximum significant bits of the w quantized coefficients introduced first on the line (y-1), which represents the line before the line y, indicated by the variable y stored in the determination unit 161 strings like Binit = 0, and saves that value.

For example, in the case where the row (y − 1) is the row L1 shown in FIG. 20, the value of the variable Binit denoting the maximum significant digit w of the quantized coefficients introduced first on the row (y − 1) is the maximum significant digit of w quantized coefficients from the left edge of line L1 in the drawing, that is, w quantized coefficients at positions (0, 0), (1, 0), ..., (w-1, 0). In addition, in the case where the variable y stored in the string definition module 161 is equal to y = 0, the string (y − 1) does not exist, as a result, the value of the variable Binit is Binit = 0.

In step S143, the row determination module 161 determines whether 0 (absolute values) of all quantized coefficients of the row y, denoted by the variable y stored in it, are equal. For example, in the case where the string y is the string L1 shown in FIG. 20, the string determination module 161 determines that all the quantized coefficients are 0, in the case where all the quantized coefficients in the positions (x, y) on the string L1 are equal 0.

In the case where it will be determined in step S143 that all the quantized coefficients are 0, the row determination module 161 generates information indicating that all the quantized coefficients are 0, and transmits it to the efficiency encoding module 62 and the maximum significant discharge calculating module 163, and the processing flow proceeds to step S144.

In step S144, based on the information indicating that all the quantized coefficients are 0, the efficiency coding unit 162 outputs (transmits) a code 0 indicating that all the quantized coefficients in the row to be encoded are 0 to the code connecting unit 169 . The code connection module 169 receives the code 0 transmitted from the efficiency encoding module 162 and outputs it without modification as the code obtained by encoding the quantized coefficients of the string y.

In step S145, the maximum significant digit calculation unit 163 sets the value of the variable Binit stored in it to Binit = 0, and updates the variable Binit based on information indicating that all the quantized coefficients are 0 from the row determination unit 161.

In step S146, the line determination module 161 determines whether or not the raw lines are in the lines of the encoded subband. That is, row determination module 161 determines whether quantized coefficients have been encoded for all rows of the encoded subband. For example, in the case of encoding the quantized subband coefficients shown in FIG. 20, the row determination module 161 determines that there are no raw rows at the point at which the quantized coefficients were encoded at all positions on the row L1-L6.

In the case where it is determined in step S146 that there is an unprocessed row, the row determination module 161 transfers the processing flow to step S147 to encode the quantized coefficients at each position in the next row, that is, in the row (y + 1).

In step S147, the line determination module 161 sequentially increments the variable y denoting the stored line as y = y + 1, returns the processing flow to step S143, and again performs the subsequent processing described above.

Conversely, in the case where it is determined in step S146 that there are no more raw rows, the quantized coefficients for all the rows making up the subband have been encoded, and thus, the row determination module 161 ends the entropy encoding processing, the stream returns to step S 113 of FIG. 23, and encoding processing ends.

In addition, in the case where it is determined in step S143 in FIG. 24 that not all of the quantized row coefficients y are 0 (there is a quantized coefficient that is not 0), the row determination module 161 generates information indicating that not all quantized coefficients are equal to 0 (there is a quantized coefficient that is not equal to 0), transfers it to the efficiency encoding module 161 and the maximum significant discharge calculation module 163, and transfers the processing flow to step S148.

In step S148, based on the information from the row determination module 161 indicating that not all quantized coefficients are 0, the efficiency encoding module 162 outputs (transmits) code 1 indicating that not all quantized row coefficients intended for encoding are 0, to code connection module 169.

In step S149, based on information indicating that all the quantized coefficients are 0 coming from the row determination module 161, the maximum significant digit calculating section 163 sets the value of the x variable indicating the x coordinate of the position on the row of the first quantized coefficient that is intended to be input in w quantized coefficients, which should now be encoded, equal to x = 0, and saves the variable x.

For example, in the case where the string y is the string L1 shown in FIG. 20, the value of the variable x, which is stored by the maximum significant digit calculating unit 163, denotes w continuous positions (x, 0), (x + 1, 0), ..., (x + w-1, 0), on the encoded line L1, the x coordinate in the farthest upper left position, (x, 0).

In addition, in step S149, the maximum significant digit calculation unit 163 sets the value of the variable B, which denotes the maximum significant bits w of the quantized coefficients encoded in the previous time, to B = Binit, and stores the variable B. That is, the maximum significant digit calculation module 163 updates the value of variable B as the value of the Binit variable stored in it, and stores the value of the updated variable B.

After updating the value of variable B, the maximum significant discharge calculation unit 163 transmits information indicating the updated variable B (maximum significant discharge) to the efficiency encoding module 164 and the significant discharge allocation module 165. In addition, each of the efficiency encoding module 164 and the significant bit allocation module 165 stores a value of the variable B supplied from the maximum significant bit calculation module 163.

In step S150, the entropy encoding unit 123 performs encoding processing of the set w. Although the encoding processing of the set w will be described in detail below, in the encoding processing of the set w, the entropy encoding unit 123 encodes w continuous quantized coefficients on the string y, denoted by the variable y, stored in the string definition module 161.

Further, using the position on the string y of the identified variable y stored in the string definition module 161 and the variable x stored in the maximum significant digit calculating module 163 as (x, y), w of consecutive positions on the string y is consecutive positions (x, y), (x + 1, y), ..., (x + w-1, y) on the string y. That is, in the encoding processing of the set w, the entropy encoding unit 123 encodes the quantized coefficients in each of the positions

(x, y), (x + 1, y), ..., (x + w-1, y).

In step S151, the maximum significant digit calculating unit 163 determines whether or not the raw quantized coefficients remain on the string y. That is, the module for calculating the maximum significant digit determines whether all the quantized coefficients have been encoded at the positions on the string y, denoted by the variable y, stored in the string definition module 161.

In the case where it is determined in step S151 that there are unprocessed quantized coefficients on row y, the maximum significant digit calculating section 163 transfers the processing flow to step S152 to encode the next w quantized coefficients.

In step S152, the maximum significant digit calculation unit 163 takes the value x stored in it as x = x + w and returns the processing flow to step S150. In accordance with this, the quantized coefficients for each of the positions (x + w, y), (x + w + 1, y), ..., (x + 2w-1, y) on the string y are encoded during processing on the following step S150.

In addition, in the case where it is determined in step S151 that there are no more unprocessed quantized coefficients on the string y, the quantized coefficients at all positions on the string y were encoded, as a result of which the maximum significant digit calculating module 163 returns the processing flow to step S146, and subsequent processing is performed.

Thus, entropy encoding unit 123 encodes the quantized coefficients at each of the positions of the subband, each at a given number, in raster scan order.

Thus, by encoding the quantized coefficients in each position of the subband at a given number each, in raster scan order, the input quantized coefficients can be processed in the order in which they are entered, and as a result, the delay associated with the encoding of the quantized coefficients can be reduced.

Next, encoding processing of the set w corresponding to the processing performed in step S50 shown in FIG. 24 will be described.

In step S181, the maximum significant discharge calculation unit 163 takes the position of the string y identified by the variable x stored in it as (x, y), and w of the continuous positions (x, y), (x + 1, y), ..., (x + w-1, y), takes the significant bits of the quantized coefficients having the largest absolute value, as the value of the variable Bnew, denoting the maximum significant bits w of the quantized coefficients that should now be encoded, and store this variable value of Bnew.

In addition, the maximum significant digit calculation module 163 transmits the obtained maximum significant digit w of the quantized coefficients, that is, the value of the Bnew variable, to the efficiency encoding module 164 and the significant digit extraction module 165.

For example, in the case where the quantized coefficients in w consecutive positions are the quantized coefficients “-0101”, “+0011”, “-0110” and “+0010” shown in FIG. 21, of these quantized coefficients, the quantized coefficient c the largest absolute value is “-0110”, and its significant digit is “3”, which is the digit in which 1 of the highest order is located for “-0110", as a result of which the value of the variable Bnew is set to 3.

In step S182, the efficiency encoding module 164 determines whether or not B = Bnew is equal. That is, the efficiency encoding module 164 determines whether or not the value of the variable B denoting the largest significant bit among the w quantized coefficients encoded the previous time is the same as the value of the variable Bnew indicating the largest significant bit w of the quantized coefficients that should now be encoded, which were transmitted from the module 163 calculating the maximum significant discharge.

In the case where it is determined in step S182 that B = Bnew, the efficiency encoding module 164 transfers the processing flow to step S183 and outputs a code 0 to the code connecting module 169, indicating the maximum significant bit w of quantized coefficients that should be encoded from now on. After the code 0 denoting the maximum significant bit is output, the efficiency encoding module 164 skips the processing from step S 184 to step S188, and transfers the processing flow to step S189.

Conversely, in the case where it is determined in step S182 that condition B = Bnew is not satisfied, the efficiency encoding unit 164 transfers the processing flow to step S184 and (since the maximum significant bit has changed), outputs code 1 to the code connection unit 169, which indicates that the maximum significant digit has changed.

In step S185, the efficiency encoding module 164 obtains integers n and m that satisfy the following expression (8).

Here, the symbol "∧" in expression (8) represents the exponent. Accordingly, (-1) ^∧m represents (1) a degree m.

For example, in the case when Bnew = 3 and B = 0, n = 2 and m = 0, they are obtained as the values of n and m, which satisfy the expression (8). When comparing the variable Bnew with variable B, the greater the difference between the absolute value of the variable Bnew and the absolute value of variable B, the greater the value of n in expression (8). In addition, in the case when the value of the variable Bnew is greater than the value of the variable B, the value of m is 0, and, conversely, in the case where the value of the variable Bnew is less than the value of the variable B, the value of m is 1. Accordingly, the value of m in expression (8) can be said to indicate whether the maximum significant digit has increased or decreased.

In step S186, the efficiency encoding module 164 outputs to the code connection module 169 a value m that satisfies expression (8) as a 1-bit code as a code indicating whether the maximum significant digit has increased or decreased. For example, in the case where the value of m that satisfies expression (8) is 0, the efficiency encoding module 164 outputs a code 0 indicating that the maximum significant digit has increased.

In step S187, the efficiency encoding module 164 outputs such a number of zeros as an n value satisfying expression (8), followed by 1 unit, as a code indicating the number of changes in the maximum significant digit, to the code connection module 169. That is, the efficiency encoding module 164 outputs n 0 and one 1 as a code indicating the number of changes in the maximum significant digit.

For example, in the case where the value of n that satisfies expression (8) is 2, the efficiency coding unit 164 outputs the value “001” as a code indicating the amount of change in the maximum significant digit to the code connecting unit 169.

Thus, the efficiency encoding module 164 outputs to the code connection module 169, a code indicating that the maximum significant digit has changed, a code indicating whether the maximum significant digit has increased or decreased, and a code indicating the magnitude of the change in the maximum significant digit, as a code indicating the maximum significant bit of w quantized coefficients that must be encoded, starting from this moment.

In step S188, the maximum significant discharge calculation unit 163 sets the value of the stored variable B to B = Bnew, and transfers the processing flow to step S189. That is, the module for calculating the maximum significant discharge updates the value of the stored variable B so that it is equal to the value of the variable Bnew stored in it. In addition, the efficiency encoding module 164 and the significant bit allocation module 165 also set the value of the stored variable B to B = Bnew.

After the variable B is set to B = Bnew in step S188, or the code indicating the maximum significant digit of the quantized coefficients is displayed in step S183, in step S189, the maximum significant digit calculation module 163 sets the value of the stored variable Binit to Binit = B , in the case when the value of the stored variable x is 0.

That is, in the case where the value of the stored variable x is 0, the maximum significant digit calculation unit 163 updates the Binit variable by taking into account the value of the Binit variable indicating the maximum significant digit w of the quantized coefficients introduced first on the row (y-1) stored in it, the value of the variable B, denoting the maximum significant bit w of quantized coefficients encoded the previous time.

Thus, in the case when the variable x = 0, the value of the variable Binit is set to Binit = B, as a result of which the encoding of the quantized coefficients can be performed for w quantized coefficients, starting from x = 0 in the next row (for example, the row (у + 1 )), using the correlation relationship with the maximum significant digit w of the quantized coefficients starting from x = 0 in the previous row (for example, row y).

In step S190, the significant digit extraction unit 165 receives the predetermined variable i and changes the variable i from 0 to (w-1) to extract significant bits of the quantized coefficients from the quantized coefficients at positions (x-i, y) on the string y. The significant bit allocation unit 165 transmits the extracted (data) significant bits of the quantized coefficients to the efficiency encoding unit 166 and the character extraction unit 167. In addition, the efficiency encoding module 166 outputs codes denoting the absolute values of w quantized coefficients to the code connection module 169 based on the significant bits transmitted from the significant bit allocation module 165 (coding of the significant bits).

Now, the value of x at the positions (x + i, y) is the value of the variable x stored in the maximum significant discharge calculation unit 163. For example, in the case where the value of the variable x stored in the maximum significant discharge calculation unit 163 is 0, the value of the variable B stored in the significant discharge allocation unit 165 is 3, and in addition, the quantization unit 122 passes to the allocation unit 165 significant digit w (four) of the quantized coefficients "-0101", "+0011", "-0110", "+0010" shown in Fig.21, corresponding to the quantized coefficients in each of the positions (X + i, y) (0 <i <3), i.e., positions (0, y), (1, y), (2, y) and (3, y), the significant digit highlighting unit 165 extracts the significant digit of the quantized coefficients.

In this case, the value of the variable B stored in the significant digit extraction unit 165 is 3, and the significant digits are 3 digits, as a result of which the significant digit extracting module 165 is extracted from the quantized coefficient “-0101”, which corresponds to the position (x, y ), the value "101", which represents the three digits of the lowest order.

In the same way, the significant digit extraction module 165 sequentially extracts from the quantized coefficients “-0011”, “-0110” and “0010” in position (x + 1, y), in position (x-2, y) and in position ( x + 3, y), the values are "011", "110" and "010", which are three numbers for its lowest order. In accordance with this, (code) significant bits "101", "011", "110" and "010" among the quantized coefficients "-0101", "-0011", "-0110" and "+0010" are output to the module Efficiency coding 166 and character extraction module 167. The efficiency encoding module 166 encodes the code "101", "011", "110" and "010" transmitted from the significant bit allocation module 165 and outputs the code "101011110010" indicating the absolute value w of the quantized coefficients to the code connection module 169.

In step S191, the character extraction module 167 receives the predetermined value i and the value of the variable i from 0 to (w-1), so that it extracts the signs of the quantized coefficients from the quantized coefficients at positions (x + i, y) on the string y, in those quantized coefficients, the absolute value of which is not equal to 0, transmitted from the quantization module 122, and transmits the selected (data) characters to the efficiency encoding module 168. The efficiency encoding module 168 encodes the characters coming from the character extraction module 167, and outputs a code indicating the signs of the quantized coefficients resulting in the code connecting module 169.

After entering the code indicating the signs of the quantized coefficients from the efficiency encoding module 168, the code connection module 169 connects a code indicating whether all the quantized row coefficients are 0, a code indicating the maximum significant bits of the quantized coefficients, a code indicating the absolute value of the quantized coefficients, and code denoting the signs of the quantized coefficients transmitted from the efficiency encoding module 162, the efficiency encoding module 164, the efficiency decoding module 166, and the efficiency encoding module 168, respectively, output coupled code as an encoded image, ends the encoding processing set w, the flow returns to step S150, as shown in Figure 24, and performs the processing from step S151, and beyond.

Now, the value of x at the positions (x + i, y) is the value of the variable x stored in the maximum significant digit calculation unit 163. For example, in the case where the value of the variable x stored in the maximum significant digit calculation unit 163 is 0, and the quantization unit 122 transmits w (four) quantized coefficients "-0101", "+0011", "-0110", "+0010 "shown in Fig.21, corresponding to the quantized coefficients in each of the positions (X + i, y) (0 <i <3), that is, in the positions (0, y), (1, y), (2, y) and (3, y) none of the “-0101”, “+0011”, “-0110” and “+0010” is 0, as a result of which the character extraction module 167 extracts characters from these quantized coefficients.

In this case, the character extraction module 167 extracts the “-” sign of the quantized coefficient from the quantized coefficient “-0101” corresponding to the position (x, y).

In the same way, the sign extraction module 167 sequentially extracts the signs “+”, “-”, “+” of the quantized coefficients from the quantized coefficients “+0011”, “-0110”, “+0010” corresponding to the positions (x + 1, y ), (x + 2, y) and (x + 3, y). Thus, the signs "-", "+", "-", "+" of the quantized coefficients "-0101", "+0011", "-0110", "+0010" are output from the character extraction module 167 to the encoding module 168 Efficiency. The efficiency encoding module 168 encodes the signs “-”, “+”, “-”, and “+” of the quantized coefficients transmitted from the sign extraction module 167.

For example, after entering the “-” sign, the efficiency encoding module 168 outputs code 1 and after entering the “+” sign displays code 0, as a result of encoding the entered characters. In this case, the signs “-”, “+”, “-” and “+” of the quantized coefficients are input to the efficiency encoding module 168, as a result of which the efficiency encoding module 168 outputs the code “1010” consisting of the code “1”, “0 "," 1 "," 0 ", as a code denoting the signs of the quantized coefficients, to the code connection module 169.

Thus, entropy encoding module 123 encodes the quantized subband coefficients in the form of sets of predetermined modules and outputs a code indicating the maximum significant bits of the quantized coefficients, a code indicating the absolute values of the quantized coefficients, and a code indicating the signs of the quantized coefficients.

Thus, by encoding the quantized subband coefficients in the form of sets of predetermined modules, the need for multiple processing in the bit plane based on a plurality of coding paths is eliminated, in contrast to the case of coding an image, for example, in JPRG 2000, and coding with a variable code length is also performed in as a result, the amount of processing during encoding can be significantly reduced. Accordingly, the encoding of the image can be performed at a higher speed, and an encoding device for real-time encoding of the high-resolution image can be implemented without cost.

Furthermore, there is no need to explicitly encode the code length in the image encoding apparatus 111 in the case of image encoding, whereby the amount of code can be reduced, and there is no need to manage information regarding the code length.

It should be noted that although the description for w quantized coefficients was described above, the significant digit of the quantized coefficient having the largest absolute value is used as the value of the Bnew variable denoting the maximum significant digit, however, it is sufficient that the value of the Bnew variable is a value equal to or greater than the significant digit of the quantized coefficient having the largest absolute value of w among the quantized coefficients. If the value of the Bnew variable is greater, the number of codes increases for the code indicating the absolute value of the quantized coefficients, but the amount of increase in the code for the code indicating the absolute value of the quantized coefficients can be reduced by setting the value of the Bnew variable to a value equal to or greater than the significant digit quantized coefficient having the largest absolute value.

The following is a description of an image decoding apparatus for decoding an image encoded by the image encoding apparatus 111.

26 is a block diagram illustrating an example configuration of an image decoding apparatus.

The image decoding apparatus 211 is configured as an entropy decoding module 221, an inverse quantization module 222, and an inverse wavelet transform module 223, wherein the encoded image (data) is input to the entropy decoding module 221.

Entropy decoding unit 221 performs entropy decoding of a code that is an input encoded image and transmits the obtained quantized coefficients, thus obtained in inverse quantization unit 222.

The inverse quantization module 222 inverts the quantized coefficients transmitted from the entropy decoding module 221 and transmits the wavelet coefficients for each subband obtained by inverse quantization to the wavelet transform module 223.

The inverse wavelet transform module 223 exposes the wavelet coefficients for each subband transmitted from the inverse quantization module 222 to the inverse wavelet transform, and outputs the resulting image as a decoded image.

In addition, the entropy decoding unit 221 of the image decoding apparatus 211 that performs such processing is configured as shown in more detail, for example, in FIG. 27.

In more detail, the entropy decoding module 221 has a code separation module 251, a string determination module 252, a generation module 253, an efficiency decoding module 254, an efficiency decoding module 255, an efficiency decoding module 256, a quantized coefficient synthesis module 257, and a switching module 258.

The code splitting unit 251 separates a code that is an encoded image that was input based on information transmitted from each of the line determination unit 252, decoding unit 254, efficiency, efficiency decoding unit 255, and efficiency decoding unit 256, and transmits the divided code to a predetermined lengths to a string determination module 252, an efficiency decoding module 254, an efficiency decoding module 255, or an efficiency decoding module 256.

That is, code splitting module 251 splits the input code into a code that indicates whether all quantized coefficients of a single encoded string are 0 or not, a code indicating the maximum significant digit w of encoded quantized coefficients, a code indicating absolute values of w encoded quantized coefficients, and a code denoting the signs of the encoded quantized coefficients, and transmits them, respectively, to each of the line determination module 252, decoding module 254, efficiency, efficiency decoding module 255 and decoding module 256 efficiency.

Line determination module 252 determines whether or not all quantized coefficients of one row of the encoded subband are equal, based on the code transmitted from code division module 251, and transmits information indicating the results of its determination to code division module 251, generation module 253 and module 254 decoding efficiency.

The generating unit 253 generates a code indicating one row corresponding to the quantized coefficients that are 0 based on the information indicating the determination results from the row determining unit 252, and transmits it to the switching unit 258.

The efficiency decoding module 254 decodes the code transmitted from the code separation module 251 denoting the maximum significant bit w of the encoded quantized coefficients, obtains the maximum significant bit w of the encoded quantized coefficients, and transmits information indicating the received maximum significant bit to the code separation module 251, efficiency decoding module 255 and a quantized coefficient synthesis module 257.

The efficiency decoding module 255 decodes a code that denotes the absolute values of the quantized coefficients transmitted to the code separation module 251 based on information indicating the maximum significant bit from the efficiency decoding module 254 and transmits (data) significant bits w of the quantized coefficients obtained by it, to the efficiency decoding module 256 and to the quantized coefficient synthesis module 257. In addition, the efficiency decoding module 255 transmits information indicating the results of code decoding, which indicates the absolute values of the quantized coefficients, to the code separation module 251.

Based on the significant bits of the quantized coefficients transmitted from the efficiency decoding module 255, the efficiency decoding module 256 decodes a code that denotes the signs of the quantized coefficients transmitted from the code division module 251 and transmits (data) the signs of the quantized coefficients thus obtained in the module 257 synthesis quantized coefficient. In addition, the efficiency decoding module 256 transmits information that indicates the code decoding results indicating the signs of the quantized coefficients to the code separation module 251.

The quantized coefficient synthesis module 257 synthesizes the significant bits of the quantized coefficients transmitted from the efficiency decoding module 255 and the signs of the quantized coefficients transmitted from the efficiency decoding module 256 based on information indicating the maximum significant bit transmitted from the efficiency decoding module 254 and transmits w quantized coefficients thus obtained in the switching module 258.

The switching unit 258 outputs the quantized coefficients from the generating unit 253 or from the quantized coefficient synthesis unit 257.

On Fig shows a block diagram illustrating a more detailed configuration of the module 251 code separation.

The code separation unit 251 has a control unit 271 and a memory 272. After entering a code that is an encoded image, the control unit 271 transmits an input code to the memory 272 for temporary storage.

The control unit 271 then reads a code of a predetermined length from a code temporarily stored in the memory 272 based on information transmitted from each line determination unit 252, the efficiency decoding module 254, the efficiency decoding module 255, and the efficiency decoding module 256, and transmits to the module 252 line definitions, efficiency decoding module 254, efficiency decoding module 255 or efficiency decoding module 256.

In addition, code separation module 251 may be implemented as shown in FIG. 29, in addition to the configuration example shown in FIG. 28.

The code separation module 251 shown in FIG. 29 has a control module 291, a switch 292, and a node 293-1, a node 293-4.

After entering a code that is encoded image, the control unit 291 controls the switch 292 based on information transmitted from each of the line determination unit 252, the efficiency decoding module 254, the efficiency decoding module 255 and the efficiency decoding module 256, and passes to the determination module 252 strings, efficiency decoding module 254, efficiency decoding module 255 or efficiency decoding module 256, code of a given length.

That is, each node 293-1 - node 293-4 is connected to the line determination module 252, the efficiency decoding module 254, the efficiency decoding module 255 and the efficiency decoding module 256, respectively, and the control module 291 selects one of the node 293-1 - node 293 -4 as the code transfer destination, and controls the connection between switch 292 and the selected node.

A switch 292 connects the node selected based on the control of the control unit 291 to an input, as a result of which the code input to the code separation unit 251 is passed to the line determination unit 252, the efficiency decoding module 254, the efficiency decoding module 255 or the efficiency decoding module 256, as the destination of the code, through the switch 292 and the node connected to the switch 292.

Next, the processing performed by the image decoding apparatus 211 will be described with reference to the flowchart shown in FIG. 30. Such decoding processing begins after entering a code, which is an encoded image, into entropy decoding unit 221.

In step S231, the entropy decoding unit 221 performs entropy decoding processing, performs entropy decoding of a code that is an input encoded image, and transmits the quantized coefficients thus obtained to the inverse quantization unit 222. Although the entropy decoding processing will be described in detail below, with such entropy decoding processing, the entropy decoding unit 221 decodes the quantized coefficients in successive positions on the encoded subband line w simultaneously and transmits the decoded quantized coefficients to the inverse quantization module 222.

In step S232, the inverse quantization unit 222 quantizes inversely the quantized coefficients transmitted from the entropy decoding unit 221 and transmits the wavelet coefficients of each subband obtained by the inverse quantization to the inverse wavelet transform unit 223.

In step S233, the inverse wavelet transform module 223 submits the wavelet coefficients of each subband that were transmitted from the inverse quantization module 222 to the inverse wavelet transform, and outputs the obtained image as its result, to which decoding processing ends.

Thus, the image decoding apparatus 211 decodes and outputs the encoded image.

Next, entropy decoding processing corresponding to the processing performed in step S231 of FIG. 30 will be described with reference to the flowchart shown in FIG.

In step S261, the string determination module 252 receives a variable y denoting a subband string, which will now be decoded as y = 0, and stores it.

In step S262, the efficiency decoding module 254 receives a variable Binit denoting the maximum significant bits w of the quantized coefficients introduced first on the line (y-1), which is one line next to the line y, designated by the variable y, stored in the line definition unit 252 like Binit = 0, and saves it.

For example, in the case where the row (y − 1) is the row L1 shown in FIG. 20, the value of the variable Binit denoting the maximum significant digit w of the quantized coefficients introduced first on the row (y − 1) is the maximum significant digit w quantized coefficients from the left edge of the row L1 in the drawing. In addition, in the case where the variable y stored in the string definition module 152 is y = 0, the string (y − 1) does not exist, as a result of which the value of the variable Binit is Binit = 0.

In addition, in step S262, the code splitting unit 251 transmits to the string determining unit 252 a first 1-bit input code code as a code indicating whether all quantized row coefficients intended for decoding are now equal to 0.

In step S263, the string determination module 252 determines whether 0 1-bit code read into (transmitted) from the code separation module 251 generates information indicating the determination result and passes to the generation module 253, the efficiency decoding module 254, and the efficiency module 251 code separation.

In the case where it is determined in step S263 that the code is 0, this means that all the quantized row coefficients y are 0, therefore, the row determination unit 252 transfers the processing flow to step S264. In step S264, the generating unit 253 takes all the quantized coefficients on the string y equal to 0 based on information indicating the result of the determination from the string determining unit 252. The generating unit 253 then generates a code indicating the quantized coefficients of the string y and passes them to the switching unit 258.

For example, in the case where one quantized coefficient is represented by four digits, as shown in Fig. 21, and there are five quantized coefficients on one row, the generating unit 253 generates 20 (= 4 × 5) zeros as a code indicating the quantized row coefficients y , and transfers them to the switching module 258. The switching unit 258 outputs 20 continuously next 0 transmitted from the generating unit 253 as a code indicating the quantized coefficients of one row to the inverse quantization unit 222.

In step S265, the efficiency decoding module 254 sets the value of the Binit variable stored in it as Binit = 0 based on the information indicating the determination results from the row determination module 252, and updates the Binit variable.

In step S266, the line determination module 252 determines whether or not raw lines are among the lines of the decoded subband. That is, row determination module 252 determines whether quantized coefficients at positions on all rows of a decoded subband have been decoded.

In the case where it is determined in step S266 that there is an unprocessed string, the string determining unit 252 transfers the processing flow to step S267 to decode the quantized coefficients at each position on the next line (y + 1) of the string y, denoted by the variable y stored in it most.

At step S267, the line determination module 252 increments the variable y denoting the stored line y = y + 1, returns the stream to step S263, and performs subsequent processing.

Conversely, in the case where it is determined in step S266 that there are no more raw lines, the quantized coefficients for all the lines making up the subband have been decoded, and as a result, the line determination unit 252 ends the entropy decoding processing, the stream returns to step S231 by 30, and processing is performed starting from step S232.

In addition, in the case where it is determined in step S263 of FIG. 31 that the code is not 0, the line determining unit 252 transfers the processing flow to step S268. In step S268, based on the information indicating the determination results from the row determination module 252, the efficiency decoding module 254 sets the value of the variable x denoting the x coordinate of the position on the row of the first quantized coefficient, which will be entered into w quantized coefficients, which will be now decoded equal to x = 0, and saves this variable x.

In addition, in step S268, the efficiency decoding module 254 receives the value of the variable B denoting the maximum significant bit w of the quantized coefficients decoded the previous time as B = Binit and stores this variable B. That is, the efficiency decoding module 254 updates the variable B using the value of variable B, as the value of the stored variable Binit and stores the value of the updated variable B.

In addition, in step S268, the code splitting unit 251 transmits the following 1-bit input code code to the efficiency decoding module 254 as a code indicating whether the maximum significant bit w of the quantized coefficients that should be encoded from a given point in time has been changed , based on the information indicating the determination results from the row determination module 252.

In step S269, the entropy decoding unit 221 performs set decoding processing. Although the decoding processing of the set w will be described in detail below, in the decoding processing of the set w, the entropy decoding unit 221 decodes w continuous quantized coefficients on the string y, denoted by the variable y, stored in the string definition module 252.

In step S270, the efficiency decoding module 254 determines whether there are unprocessed quantized coefficients on the string y. That is, the efficiency decoding module 254 determines whether all of the quantized coefficients at the positions of the string y indicated by the variable y stored in the string definition module 252 have been decoded.

In the case where it is determined in step S270 that there are raw quantized coefficients on line y, the efficiency decoding unit 254 transfers the processing flow to step S271 to decode the next w quantized coefficients.

In step S271, the efficiency decoding unit 254 receives the value of the variable x stored in it as x = x + w, and returns the processing flow to step S269. In accordance with this, the quantized coefficients in each of the positions (x + w, y), (x + w + 1, y), ..., (x + 2w-1, y) on the string y are decoded during processing performed on the next step is S269.

In addition, in the case where it is determined in step S270 that there are no more unprocessed quantized coefficients on the string y, the quantized coefficients in all positions on the string y were decoded, as a result of which the efficiency decoding module 254 returns the processing flow to step S266, and the subsequent treatment.

Thus, entropy decoding unit 221 decodes the quantized coefficients at each position of the subband by a predetermined number each, in raster scan order.

Thus, by decoding the quantized coefficients in each of the subband positions by a predetermined number each in raster scan order, the input quantized coefficients can be processed in the order in which they are input, and the delay due to the encoding of the quantized coefficient can be reduced.

Next, decoding processing of the set w corresponding to the processing performed in step S269 of FIG. 31 will be described with reference to the flowchart of FIG. 32.

As described above, in step S268 of FIG. 31, a 1-bit code is transmitted to the code decoding module 251 — the efficiency decoding module 254 indicating whether the maximum significant bit w of the quantized coefficients for decoding has changed.

In step S311 of FIG. 32, the efficiency decoding module 254 determines whether the read (transmitted) 1-bit code is 0.

In the case where it is determined in step S311 that the read code is 0, the maximum significant bit has not changed, as a result of which the efficiency decoding module 254 generates information meaning that the maximum significant bit has not changed and transfers it to the code separation module 251, the module The efficiency decoding 255, and the quantized coefficient synthesis module 257, skips the processing in step S312-S314, and transfers the processing flow to step S315.

That is, in the case where the code indicating whether the maximum significant digit has changed or not is 0, the input character following the 1-bit code equal to 0, indicating whether the maximum significant digit has changed, is not a code indicating that the maximum significant digit the bit increased or decreased, and a code indicating the magnitude of the change in the maximum significant discharge, but rather a code indicating the absolute value of the quantized coefficients, resulting in steps S312 - step S314, which indicate the processing code indicating, whether increased or decreased if the maximum significant digit and the code indicating the amount of change of the maximum significant digit is passed.

Conversely, in the case where it is determined in step S311 that the read code is not 0, the maximum significant bit has changed, as a result of which the efficiency decoding unit 254 transfers the processing flow to step S312, reads the 1-bit code from the code separation unit 251 and stores its value as a given variable m.

In step S313, the efficiency decoding module 254 reads the code from the code splitting module 521 until the code becomes 1 (until the code 1 is counted), and stores the number of codes equal to 0 read so far as variable n. For example, in the case when the third code that the efficiency decoding module 254 read from the code separation module 251 is 1, that is, in the case where the efficiency decoding module 254 read the code “001”, the number of codes 0 is read until the module The efficiency decoding decoding 254 counted 1, equal to 2, as a result of which the efficiency decoding decoding module 254 stores 2, which is the number of codes read 0, and the value of the variable n.

In step S314, the efficiency decoding module 254 obtains a value B denoting the maximum significant bits from the following expression (9), and stores the value of the obtained variable B.

Now, the left side of the expression (9) represents the value of the variable B, which must be obtained again, while the position of the right side shows the value of the stored variable B. In addition, the symbol " ^∧ " in the expression (9) represents the exponent. Accordingly, (-1) ^∧ m is (-1) in m-th degree.

The efficiency decoding module 254 calculates by expression (9) and changes the stored variable B based on the stored variable B, variable m, and variable n. After updating the variable B indicating the maximum significant bit, the efficiency decoding module 254 generates information indicating the updated maximum significant bit and passes it to the code separation module 251, the efficiency decoding module 255 and the quantized coefficient synthesis module 257.

After receiving a new maximum significant digit in step S314 or determining in step S311 that the 1-bit code that was read is 0, the efficiency decoding unit 254 transfers the processing flow to step S315, and in the case where the value of the stored variable x is 0 , sets the value of the saved Binit variable to Binit = B.

That is, in the case where the value of the stored variable x is equal to 0, the efficiency decoding module 254 takes the stored value of the variable Binit, denoting the maximum significant digit w of the quantized coefficients introduced first on the line (y-1), as the value of the variables denoting the maximum significant bit w quantized coefficients that must be decoded from this point in time, and updates the Binit variable.

Thus, in the case when the variable x = 0, the value of the variable Binit is set to Binit = B, as a result of which the encoding of the quantized coefficients can be performed for w quantized coefficients, starting from x = 0 in the next row (for example, the row (у + 1)), using the correlation relationship with the maximum significant digit w of the quantized coefficients, starting from x = 0 in the previous row (for example, row y).

In step S316, the efficiency decoding module 255 receives the given variable i and changes the variable i from 0 to (w-1), as a result of which the code is read from the code splitting module 251, in increments of B bits, and transmits (outputs) B bits of the read code to an efficiency decoding module 256, and a quantized coefficient synthesis module 257, as a code indicating significant bits of the quantized coefficients at positions (x + i, y) on the string y. In addition, the efficiency decoding module 255 generates information indicating significant bits of the quantized coefficients, and transmits it to the code separation module 251.

Now the value of x at positions (x + i, y) is the value of the variable x stored in the efficiency decoding unit 254. For example, in the case where the value of the variable x stored in the efficiency decoding unit 254 is 0 and the value of the variable B stored in the efficiency decoding unit 255 is 3, the efficiency decoding unit 255 reads 3 bits of code from the code splitting unit 251, with the variable i = 0, and outputs a code of 3 read bits, as a significant bit of the quantized coefficient in position (0, y).

In the same way, the efficiency decoding module 255 reads the other 3 bits of code from the code separation module 251, with the variable i = 1, and outputs the code 3 of the read bits as a significant bit of the quantized coefficient in position (1, y), reads the next 3 bits of code output from the code splitting module 251 with the variable i = 2 and outputs the code of these 3 bits read as the significant bit of the quantized coefficient in position (2, y), and reads the next 3 bits of code from the code splitting module 151 with the variable i = 3, and outputs a code of 3 read bits as a significant digit quantized th coefficient in position (3, y).

In step S317, the efficiency decoding module 256 takes the given variable i and changes this variable i from 0 to (w-1) and in the case when the significant bits (absolute values) of the quantized coefficients in the positions (x + i, y) on the line y are not equal to 0, reads 1 bit of code from module 251 code separation. The efficiency decoding module 256 then decodes this read code and transmits (outputs) the resulting code to the quantized coefficient synthesis module 257 as signs of the quantized coefficients. In addition, the efficiency decoding module 256 generates information indicative of the signs of the quantized coefficients and transmits it to the code separation module 251.

Now, the value of x at positions (x + i, y) takes on the value of the variable x stored in the efficiency decoding unit 254. For example, in the case where the value of the variable x stored in the efficiency decoding module 254 is 0, and (code designation) a significant bit other than 0 is transmitted from the efficiency decoding module 255, the efficiency decoding module 256 takes the variable i = 0 and reads 1 bit of code from the code separation module 215 and, in the case where the code is 0, transmits a code denoting the “-” sign of the quantized coefficient at position (0, y) to the quantized coefficient synthesis module 257, and in the case when the code is 1, transmits a code denoting the sign “+” of the quantized coefficient in the position ii (0, y), synthesis module 257 the quantized coefficient.

In addition, in the case where the absolute value (code designation) of the significant bit transmitted from the efficiency decoding module 255 is 0, the sign of the quantized coefficient is not in position (0, y), as a result of which the efficiency decoding module 256 does not read the code from the module 251 code separation.

In the same way, in the case when the (absolute value) significant bit transmitted from the efficiency decoding module 255 is not equal to 0, the efficiency decoding module 256 takes the variable i = 1 and reads 1 bit of code from the code separation module 215, and in the case when the code is 0, it transmits the code indicating the “-” sign to the quantized coefficient synthesis module 257, and in the case where the code is 1, it transmits the code indicating the “+” sign.

In addition, in the case where the significant bit transmitted by the next one from the efficiency decoding module 255 is not equal to 0, the efficiency decoding module 256 takes the variable i = 2 and reads 1 bit of code from the code separation module 215, and in the case when the code is 0 transmits a code indicating the sign “-” to the module 257 synthesis of the quantized coefficient, and in the case when the code is 1, transmits a code indicating the sign “+”. In addition, in the case where the significant bit transmitted by the next one from the efficiency decoding module 255 is not equal to 0, the efficiency decoding module 256 takes the variable i = 3 and reads 1 bit of code from the code separation module 215, and in the case when this code is 0, transmits a code denoting the “-” sign to the quantized coefficient synthesis module 257, and when the code is 1, transmits a code denoting the “+” sign.

In step S318, the quantized coefficient synthesis module 257 synthesizes the significant bits transmitted from the efficiency decoding module 255 and the characters transmitted from the efficiency decoding module 255, outputs the resulting quantized coefficients to the inverse quantization module 222 through the switching module 258, ends the decoding processing of the set w and returns the flow to step S269 of FIG. 31, and performs processing starting from step S270.

For example, the number of bits in the absolute values of the quantized coefficients intended for output is determined in advance. In the case where the number of bits of the absolute value of the quantized coefficients intended for output, which is predetermined, is 4 bits, and the maximum significant bit indicated by the information indicating the maximum significant bit from the efficiency decoding unit 254 is 3, the significant bit is “101” transmit from the efficiency decoding module 255, and after the code indicating the sign “-” is transmitted from the efficiency decoding module 255, the quantized coefficient synthesis module 257 outputs the quantized coefficient “-010 one".

That is, the number of bits of the absolute value of the quantized coefficients to be output is 4 bits, and the significant bits are “101” (three bits), whereby the quantized coefficient synthesis module 257 receives one higher order bit of the significant bits “101”, as 0 for the 4-digit absolute value of the quantized coefficient pf "0101". In addition, the “-” sign of the quantized coefficient and the absolute value “0101” of the quantized coefficient are synthesized so that a code “-0101” is obtained, which is output as a quantized coefficient.

It should be noted that in the case where the significant bit transmitted from the efficiency decoding unit 255 is 0, the quantized coefficient synthesis unit 257 outputs an unsigned quantized coefficient. For example, in the case where the number of digits of the absolute value of the quantized coefficients intended for output, which is predetermined, is 4 bits, and the maximum significant bit according to information indicating the maximum significant bit from the efficiency decoding module 254 is 3, and the significant bit is "000 "transmit from the efficiency decoding module 255, the quantized coefficient synthesis module 257 outputs the quantized coefficient" 0000 ".

Thus, the entropy decoding unit 221 decodes the quantized subband coefficients as a plurality of predetermined modules.

Thus, decoding encoded quantized subband coefficients in the form of a plurality of predetermined modules eliminates the need for processing multiple times in the bit plane based on a plurality of coding paths, unlike the case of decoding an image using JPEG 2000, whereby decoding an image can be faster. Accordingly, an encoding device for decoding a high-resolution image in real time can be implemented without unnecessary costs.

Thus, the example of the image encoding apparatus 111 described above describes that the absolute values of the given w quantized coefficients are encoded in the order of encoding (or decoding) the absolute values of the quantized coefficients, but the encoding (or decoding) of the image can be faster due to the simultaneous parallel encoding (or decoding) w of quantized coefficients using the OKMD calculation instructions (SIMD, one instruction stream and many data streams) together with the DSP ( DSP, general purpose digital signal processor) or general purpose CPU.

Then, an example of OKMD operation instructions includes MMR (MMX, Multimedia Extension), SSE (additional set of OKMD streaming commands), for example SSE2, SSE3, used with CPUs manufactured by Intel Corporation, etc.

In the case of encoding the absolute values of the quantized coefficients using such OKMD operating instructions, the entropy encoding module 123 of the image encoding apparatus 111 is configured as, for example, shown in FIG. 33.

The entropy encoding module 123 shown in FIG. 33 is the same as the entropy encoding module 123 shown in FIG. 22 in that it includes a line determination module 161, an efficiency encoding module 162, a maximum significant discharge calculation module 163 , an efficiency encoding module 164, a significant bit allocation module 165, an efficiency encoding module 166, a character extraction module 167 and an efficiency encoding module 168, and characterized in that the buffer 401 is provided for the code connection module 169. It should be noted that the sections shown in FIG. 33, which correspond to those shown in FIG. 22, are denoted with the same reference numerals and are not described here.

In the buffer 401 of the code connection module 169, a code is temporarily stored indicating whether or not all quantized row coefficients are equal, a code indicating the maximum significant digit, a code indicating the absolute value of the quantized coefficients, and a code indicating the sign of the quantized coefficients, respectively, transmitted from the module 162 coding efficiency, module 164 coding efficiency, module 166 coding efficiency and module 168 coding efficiency.

The storage area of the buffer 401 is controlled in increments of 32 bits, and the code (data) entered into the buffer 401 is stored, being divided into a code used for processing scalar calculations, and a code used for processing vector calculations. That is, in one 32-bit storage area, the code used to process scalar calculations, or the code used to process vector calculations, is stored as a code (data) that should be temporarily stored.

Using the entropy encoding module 123 shown in FIG. 33, the absolute values of the quantized coefficients are encoded in parallel using the instructions of the OKMD operation, so that a code indicating the absolute values of the quantized coefficients is accepted as the code used for processing vector calculations, and another code is selected as a code used for scalar processing calculation.

It should be noted that in the following description, 32-bit storage areas provided for buffer 401, storage areas in which the code used to process the scalar calculation is stored will be called scalar areas, and storage areas in which the code used to process vector calculations will be called vector areas.

Next, the entropy encoding performed by the entropy encoding unit 123 shown in FIG. 33 will be described with reference to FIG. 34.

For example, suppose that the 12 quantized coefficients are "-0101", "+0011", "-0110", "+0010", "+0011", "+0110", "0000", "-0011", "+1101 "," -0100 "," +0111 "and" -1010 "shown in Fig. 21 are entered into the entropy encoding module 123, as shown in the upper left part of Fig. 34.

In the same way, as described with reference to FIG. 21, a code “1” is transmitted to the code connection module 169 to the entropy encoding module 123, indicating whether all quantized row coefficients that are intended to be encoded are equal to 0 and the code “10001”, denoting the maximum significant bits of the first four input quantized coefficients "-0101", "+0011", "-0110" and "+0010".

Then, the code "110001", consisting of the code "1", indicating whether all quantized row coefficients intended for encoding are equal to 0 or not, and the code "10001", indicating the maximum significant bits of the quantized coefficients, are stored in the 32-bit scalar region, provided in the buffer 401 of the code connection module 169, as indicated by arrow A11.

In the example shown in FIG. 34, the code intended to be stored in the scalar region is stored in the direction from left to right in the diagram in the order of the higher order bits. After the code is saved in all one scalar region, that is, after 32 bits of code are stored in one scalar region, one new scalar region is provided in buffer 401 and the code used in the processing of scalar calculations is sequentially stored in the newly provided scalar region.

After the code "110001" consisting of the code "1" indicating whether or not all the quantized coefficients of the string to be encoded are equal to, and the code "10001" indicating the maximum significant bits of the quantized coefficients will be stored in the scalar region, then the entropy coding unit 123 stores in the vector domain a code equivalent to the maximum significant bits representing the absolute values of each of the first w (four) quantized coefficients entered, “-0101”, “+0011”, “-0110” and “+0010” simultaneously ( lined up in parallel).

The maximum significant bits of the quantized coefficients “-0101”, “+0011”, “-0110” are, as described with reference to FIG. 21, “3”, with the result that the code representing the absolute value of the four quantized coefficients is “ 101 "," 011 "," 110 "and" 010 ", respectively, and as shown by arrow A12, the codes" 101 "," 011 "," 110 "and" 010 "representing the absolute value of the quantized coefficients are stored arranged in one vector area provided in buffer 401.

Now the vector region is further divided into four 8-bit regions, and in each of the four regions of the vector region a code is stored representing the absolute values of the four quantized coefficients of the same length (bit length) from left to right in the drawing, in order, starting from the bits of the higher order.

In the vector region shown by arrow A12 in the drawing, the code "101" indicating the absolute value of the quantized coefficient is stored in the left direction in the 8-bit region located on the left side, the code "011" is stored in the left direction in the 8-bit region, second from the left side, the code "110" is stored in the left direction in the area of the second from the right side and the code "010" is stored in the left direction in the area located in the right-most position.

In addition, in the same manner as in the case of the scalar region, in the vector region, after saving the code, in the whole one vector region, that is, after saving 32 bits of code in one vector region, one new vector region is provided in the buffer 401, and the code used in the processing of vector calculations is sequentially stored in the newly provided scalar region.

After the code indicating the absolute values of the quantized coefficients "-0101", "+0011", "-0110" and "+0010" is stored in the vector domain, the entropy coding unit 123 stores a code indicating the signs of these four quantized coefficients in the scalar region, as shown by arrow A13.

As shown by arrow A11, the code "110001", consisting of the code "1", indicating whether all quantized row coefficients are intended to be equal to 0 or not, and the code "10001", indicating the maximum significant bits of the quantized coefficients, are already stored in the scalar areas, thus, as shown by arrow A13, the code "1010" indicating the signs of the quantized coefficients "-0101", "+0011", "-0110" and "+0010" is stored on the right side of the code "110001" already saved in the scalar area (sequentially to the right of the code "110001").

Then, after encoding the first four quantized coefficients, entropy encoding unit 123 encodes the following four quantized coefficients "+0011", "+0110", "0000" and "-0011".

First, entropy coding unit 123 compares the maximum significant bits “3” of the four quantized coefficients encoded the previous time with the maximum significant bits “3” of the next four quantized coefficients “+0011”, “+0110”, “0000” and “-0011” encoded this time, and since the maximum significant bits have not changed, saves the code "0", indicating that the maximum significant bits have not changed, as the code indicating the maximum significant bits, as shown by arrow A14, in the scalar region sequentially Ava from the already saved code "1100011010".

Then, entropy coding unit 123 stores each of the codes “011”, “110”, “000” and “011” representing the absolute value of each of the w (four) quantized coefficients “+0011”, “+0110”, “0000” and The “-0011” introduced this time into the vector region, as shown by arrow A15, simultaneously.

As shown by arrow A12, in the 8-bit region of the vector region on the left side of the drawing, in the 8-bit region, the second on the left, in the 8-bit region, the second on the right and in the 8-bit region, the codes “101 "," 011 "," 110 "and" 010 ", as a result of which the entropy coding unit 123 stores each of the codes" 011 "," 110 "," 000 "and" 011 "representing the absolute value of each of the quantized coefficients entered this time, sequentially to the right of the corresponding codes "101", "011", "110" and "010" already stored in the vector area, as shown arrow A15.

Then, entropy coding unit 123 receives a code “001” indicating the signs of the quantized coefficients, in which the absolute value is not 0 for the four quantized coefficients “+0011”, “+0110”, “0000”, and “-0011” entered this time , and saves it sequentially to the right of the code "11000110100", already saved in the scalar region.

After encoding the four quantized coefficients "+0011", "+0110", "0000" and "-0011", the entropy encoding unit 123 encodes the following four quantized coefficients "+1101", "-0100", "+0111" and " -1010 ".

First, entropy encoding module 123 compares the maximum significant bits “4” of the four quantized coefficients “+1101”, “-0100”, “+0111” and “-1010” entered this time with the value “3” of the four quantized coefficients encoded the previous time, and saves the code "101", indicating the maximum significant bits, consisting of "1", which means that the maximum significant bits have changed, "0", which means that the maximum significant bits have increased, and "1", which denotes the magnitude of the increase in maximum significant digits, as shown arrow A17 in the scalar field.

In this case, the code "11000110100001" is already stored in the scalar region, as shown by arrow A16, as a result of which the entropy coding module 123 stores the code "101", which indicates the maximum significant digits to the right of the code "11000110100001" in the drawing.

Then, after saving the code denoting the maximum significant bits of the four quantized coefficients "+1101", "-0100", "+0111" and "-1010", entropy encoding module 123 stores each of the codes "1101", "0100", " 0111 "and" 1010 ", representing the absolute value of each of them, in the vector region, as shown by arrow A18, simultaneously.

As shown by arrow A15, "101011", "011110", "110000 are already stored in the 8-bit region of the vector region in the drawing, in the second from the left on the 8-bit region, in the second from the right on the 8-bit region and the rightmost 8-bit region "and" 010011 ", respectively, as a result, each of the 8-bit area on the left, the 8-bit area on the second on the left, the 8-bit area on the second on the right and the 8-bit area on the right-most position can only store two bits of code.

Accordingly, entropy encoding module 123 provides (provides) a new vector region in buffer 401, as shown by arrow A18, stores a code of two higher order bits "11", "01", "01" and "10", for codes "1101", "0100", "0111", "1010" representing the absolute value of the quantized coefficients entered this time, sequentially on the right side of the codes "101011", "011110", "110000" and "010011", already stored in the vector area, and saves two bits of the lower order "01", "00", "11" and "10" for the codes "1101", "0100", "0111", "1010" representing the absolute value of the quantum of the coefficients entered this time, on the left side of each of the 8-bit areas, the newly provided vector area (of the two vector areas indicated by arrow A18, in the vector area at the bottom of the drawing) in the drawing, the 8-bit area second from the left , The 8-bit region is second to the right and the 8-bit region is in the rightmost position, respectively.

After storing a code indicating the absolute value of the four quantized coefficients "+1101", "-0100", "+0111" and "-1010", as shown by arrow A19, entropy encoding module 123 stores the code "0101" indicating the signs of the four quantized coefficients whose absolute value is not equal to zero, sequentially on the right side of the code "11000110100001101", already stored in the scalar region.

Thus, after the input quantized coefficients are encoded, the entropy encoding unit 123 outputs, in order, the code stored in the scalar region shown by arrow A19, the code stored to the upper vector region in the drawing of two vector regions shown by arrow A19, and the code stored in the lower vector region as an encoded image.

In this case, the code is not stored in 11 bits on the right side of the drawing from the scalar region shown by arrow A19. In addition, the code is not stored in the area of the right six bits of each of the 8-bit region on the left side, the 8-bit region second on the left, the 8-bit region second on the right and the 8-bit region in the rightmost position, in the vector region in the lower parts of two vector areas indicated by arrow A19.

In the case of outputting the code stored in scalar regions and vector regions, such as an encoded image, in the case where there are regions in which the code is not stored in scalar regions and vector regions, at the time when the encoding of the input quantized coefficients is completed, save an arbitrary code, such as, for example, the code "0", after which the code stored in scalar regions and vector regions is encoded and displayed as an image.

Accordingly, in the example shown by arrow A19, for example, the code "11000110100001101010100000000000" stored in the scalar region, the code "10101111011110011100000101001110" stored in the upper vector region in the drawing, and the code "01000000000000001100000010000000" stored in the lower vector region in order, like an encoded image. Now, arbitrary code stored in scalar regions and vector regions, in regions where the code was not saved at the moment of encoding of the quantized coefficients, will not be read during decoding, so any kind of code can be saved.

In cases of encoding the absolute values of the quantized coefficients, also using the OKMD operation instruction, after image input, the image encoding apparatus 111 performs the encoding processing described with reference to the flowchart shown in FIG. 23. Furthermore, in the entropy encoding processing of FIG. 24 corresponding to step S 113 of FIG. 23, the image encoding apparatus 111 performs the same processing as in the case without using the instructions of the OKMD operation for processing from step S141 to step S149 in FIG. 24, the processing of step S151 and the processing of step S152 (the processing is described with reference to FIG. 24), and to process the encoding of the set w corresponding to step S150, performs difference processing with a case in which the operation instructions of the OCHA are not used.

The following is a description of the encoding processing of the set w in the case of the image encoding apparatus 111, which encodes the absolute values of the quantized coefficients using the operation instructions OKMD, which is given with reference to the flowchart shown in Fig. 35. It should be noted that each processing in steps S411 to S419 corresponds to each processing performed in steps S181 to S189 shown in FIG. 25, and each of them is performed in the same way. Accordingly, their description would be redundant and therefore will not be given.

In addition, in the case of encoding the absolute values of the quantized coefficients using the instructions of the OKMD operation in the processing described with reference to FIG. 24 and FIG. 35, a code indicating whether all absolute values of the quantized coefficients of the string intended for encoding are transmitted are equal to 0 an efficiency encoding module 162 to a code connection module 169, a code indicating the maximum significant bits of the quantized coefficients transmitted from the efficiency encoding module 164 to a code connection module 169, and a code indicating the sign of quanta nnyh coefficients supplied from the encoding unit 168 Efficiency, the code in the module 169 compounds are each stored in scalar regions provided in the buffer 401 of code module 169 compound, as described with reference to Figure 34.

In step S420, the significant digit extraction module 165 simultaneously extracts the significant bits of the quantized coefficients from w successive quantized coefficients at positions (x, y), (x + 1, y), ..., (x + w-1, y) on the line y transmitted from the quantization module 122. The significant bit allocation unit 165 transmits the selected significant bits of the quantized coefficients to the efficiency encoding unit 166 and the character extraction unit 167. In addition, the efficiency encoding module 166 simultaneously outputs a code indicating the absolute values w of the quantized coefficients based on the significant bits transmitted from the significant bit allocation module 165 (coding of the significant bits) to the code connection module 169.

Now, the value of x in position (x, y) is taken as the value of the variable x stored in the maximum significant digit calculating unit 163, and the value of y is taken as the value of the variable y stored in the string determining unit 161. For example, when the significant bit extraction module 165 extracts the significant bits "101", "011", "110" and "010" as the significant bits of the quantized coefficients, the code "101", "011", "110" and "010", denoting the absolute values of the four quantized coefficients are transmitted from the efficiency encoding module 166 to the code connection module 169, so that the code connection module 169 encodes a code indicating the absolute values of the quantized coefficients transmitted to it, and, as shown by arrow A12 in Fig. 34, saves in vector area.

In step S421, the character extraction module 167 receives the given variable i and changes the variable i from 0 to (w-1) so that it extracts the signs of the quantized coefficients from the quantized coefficients at positions (x + i, y) on the row y whose quantized coefficients are are not equal to 0 transmitted from the quantization module 122, and transmits the selected (data) characters to the efficiency encoding module 168. Now, the value of x in position (x, y) is taken as the value of the variable x stored in the maximum significant digit calculating unit 163, and the value of y is taken as the value of the variable y stored in the string determining unit 161.

The efficiency encoding module 168 encodes the characters from the character extraction module 167 and outputs a code indicating the signs of the quantized coefficients obtained by it to the code connection module 169. In addition, as described with reference to FIG. 34, the code connection unit 169 stores a code indicating the signs of the quantized coefficients transmitted from the efficiency encoding unit 168 in the scalar region of the buffer 401.

After storing the code denoting the signs of the quantized coefficients in the scalar region of the buffer 401, the code combining unit 169 concatenates the codes stored in the scalar regions and vector regions of the buffer 401, as described with reference to FIG. 34, outputs the connected code as an encoded image, resulting in the encoding processing of the set w ends, and the processing returns to step S150 of FIG. 24, and processing is performed starting from step S 151.

Thus, entropy encoding unit 123 simultaneously encodes the absolute values of a given number of quantized coefficients.

When using the entropy coding method in accordance with JPEG 2000, the quantized coefficients are subjected to arithmetic coding with increments of bit planes based on a plurality of coding paths, as a result of which the specified processing for entropy coding is parallel and simultaneously difficult to implement, but when using the module 123 of entropy coding, there is no need to perform complex processing with increment of bit planes, which allows you to simultaneously encode lute values of the set of quantized coefficients.

Thus, the simultaneous encoding of the absolute values of a given number of quantized coefficients allows you to simultaneously perform many processes (in parallel), which allows you to encode images at a higher speed.

It should be noted that with respect to the processing performed in step S421, a description is given indicating that the encoding of the signs w of the quantized coefficients is performed in order, but the encoding of the signs w of the quantized coefficients can be performed simultaneously using the instructions of the OKMD operations, as in the case encoding the absolute values of the quantized coefficients. In this case, each of the codes denoting the signs w of the quantized coefficients obtained by encoding is stored in the vector region of the buffer 401, which was divided by w.

Furthermore, although a description has been made with respect to buffer 401, in which one scalar region or vector region is a 32-bit region, and so that the 32-bit regions are further divided into 8-bit regions for use, size, and the like. scalar region or vector region can be arbitrary. For example, an arrangement may be adopted in which a scalar region or vector region is set as a 128-bit region, and the 128-bit region is divided into eight 16-bit regions for use.

In addition, in the case of decoding an image encoded using the operation instruction OKMD, the module 251 code separation (Fig) device 211 decoding the image that decodes the image is made, as shown, for example, in Fig, and a code that represents a coded image, 32 bits are stored simultaneously in the storage device 272, as described with reference to Fig. 34.

In the case of reading the code from the storage device 272 and its output, the control unit 271 first receives a storage area in which the first 32 bits of the code are stored as a scalar area, and reads and outputs in order, starting from the top of the scalar area, a code indicating whether 0 or not, all absolute values of the quantized coefficients of the string intended for decoding, a code indicating the maximum significant digit of the quantized coefficients, or a code indicating the signs of the quantized coefficients.

In addition, in the case of reading a code indicating the absolute values of the quantized coefficients from the storage device 272, the control unit 271 receives a 32-bit storage area after the storage area adopted as a scalar area in the storage device 272 (accordingly, the code has not yet been read from this region), as a vector region, and reads and outputs a code indicating the absolute values of the quantized coefficients from this vector region.

It should be noted that in the case when the image is encoded, this image is encoded so that during the reading of the code indicating the absolute values of the quantized coefficients for the first time after decoding, there is always a code indicating the absolute values of the quantized coefficients (the code used to calculate the vectors) stored in the next 32-bit save region that follows the save region used as the scalar region.

In addition, the number of bits that each storage area consists of, which divide the code, which is an encoded image, and stored in the storage device 272, varies according to how many bits are set for one scalar region and a vector region, in the case of image encoding using the image encoding apparatus 111. That is, the size of each of the plurality of storage areas in the storage device 272 into which the code, which is an encoded image, is divided and stored, is the size of one scalar region and a vector region in the case of image encoding.

In the case of decoding the absolute values of the quantized coefficients using the instructions of the OKMD operation, the image decoding module 211 also performs decoding processing described with reference to the flowchart shown in FIG. 30 after entering the encoded image. In addition, the entropy decoding processing shown in FIG. 31 corresponds to the processing performed in step S231 in FIG. 30 with respect to each processing of steps S261 to S268 in FIG. 31, the processing in step S270 and the processing in step S271, with this, the image decoding apparatus 211 performs the processing in the same way as when the OKMD control command is not used (the processing described with reference to FIG. 31), and in the decoding processing of the set w corresponding to step S269, performs processing different from the processing in the case when not used OKMD management teams use it.

The following is a description of the decoding processing of the set w in the case where the image decoding apparatus 221 decodes the absolute value of the quantized coefficients using the instructions of the OKMD operation with reference to the flowchart shown in FIG.

It should be noted that the processing performed in steps S451 to step S455 corresponds to the processing performed in steps S311 to step S315 in FIG. 32, and each of them is the same. In accordance with this, their excessive description is not given here.

In addition, in the case of decoding the absolute values of the quantized coefficients using the instructions of the OKMD operation, the code, which is an image, is stored in the memory 272 of the code splitting module 251, divided into three 32-bit regions, for example, as shown by arrow A19 in FIG. 34. The line definition module 252, the efficiency decoding module 254, and the efficiency decoding module 256 each accepts from these three 32-bit regions the upper region in FIG. 34 as a scalar region and sequentially reads and decodes, starting from the top from the scalar region (left side in the drawing ), a code indicating whether or not all quantized coefficients in a row are equal, a code indicating the maximum significant digit of the quantized coefficients, and a code indicating the signs of the quantized coefficients.

In step S456, the efficiency decoding module 255 simultaneously reads w sets of B consecutive code bits from code separation unit 255 and receives each of w sets of B bits of read code as a code denoting significant bits of the quantized coefficients from w consecutive quantized coefficients in positions (x, y ), (x + 1, y), ..., (x + w-1, y) on the line y, and transfers (outputs) them to the efficiency decoding module 256 and the quantized coefficient synthesis module 257. In addition, the efficiency decoding module 255 generates information indicating significant bits of the quantized coefficients and transmits them to the code separation module 251. Now, the value of x in position (x, y) is taken as the value of the variable x stored in the efficiency decoding unit 254, and the value of y is taken as the value of the variable y stored in the line definition unit 252.

For example, if you can say that the given number w is 4, the value of variable B is 3, and the code, which is an image, is stored in the memory 272 of the code splitting module 251, by dividing it into three storage areas of 32 bits each, as shown by the arrow A19 in FIG. 34, the 32-bit storage area on the uppermost side of FIG. 34 has already been adopted as a scalar area, using a code indicating whether all quantized coefficients in a row are 0 or not and a code indicating the maximum significant bit of quantized coefficients the customers that have been read, and a code that has not yet been read from the next 32-bit storage area (the storage area is second from the top), so that the module decoding efficiency module 155 receives the second storage area from the top as a vector area, and at the same time reads the codes "101", "011", "110" and "010", denoting significant bits of the quantized coefficients at the positions (x, Y), (x + 1, y), (x + 2, y) and (x + 3, y), on the left side of the 8-bit region of the vector region in the drawing, the 8-bit region is second on the left, the 8-bit region is second on the right and the 8-bit region tee at the far right.

After transmitting the code indicating significant bits w of the quantized coefficients to the efficiency decoding module 256 and to the quantized coefficient synthesis module 257 thereafter, the processing in step S457 and the processing in step S458 are performed, but these processes are the same as the processing in step S317, and the processing performed in step S318 of FIG. 32, so a description thereof will not be given here.

Thus, entropy decoding unit 221 simultaneously decodes the absolute values of a given number of quantized coefficients.

Thus, simultaneous decoding of the absolute values of a given number of quantized coefficients allows multiple processes to be performed simultaneously (in parallel), and images can be decoded at a higher speed.

It should be noted that although the description has been made with respect to the processing performed in step S457, in which the decoding of each of the codes denoting the signs of w quantized coefficients is performed in order, a layout can be made in which the decoding of each of the codes denoting the signs of w quantized coefficients, is performed simultaneously (in parallel), using the instructions of the operation OKMD.

As described above, it is not necessary to carry out arithmetic coding of quantized coefficients in increments of bit planes based on many coding paths, in contrast to the case of encoding (or decoding) an image using conventional JPEG 2000, which allows faster encoding (or decoding) of an image, at more simple handling.

Using the conventional JPEG 2000 method, processing is performed for each bit plane based on a plurality of coding paths, so that if such processing is performed, the quantized coefficients must be accessed so many times that is equal to the number obtained by multiplying the quantized coefficients by the number of bit planes such that means the processing volume was large.

In addition, in the case of encoding an encoded image, the packetization processing cannot be started until the encoding of one image is completely completed, therefore, a delay is generated equivalent to the waiting interval for such processing, respectively. In addition, in the JPEG2000 method, quantized (encoded) coefficients corresponding to positions in a rectangular region consisting of parallel sides in the X direction and Y direction in the subband shown in FIG. 2, for example, are stored in one packet, so a delay is also generated, corresponding to the length in the Y direction of the rectangular region. Using the conventional JEPG 2000 method, such delays are generated due to encoding, which makes it difficult to provide real-time processing. Moreover, although the delay can be reduced by shortening the length in the Y direction of the rectangular region in the subband, in this case, the coding efficiency is degraded.

Conversely, when using the image encoding device 111, it is not necessary to arithmetically encode the quantized coefficients for each bit plane based on a plurality of encoding paths, as described above, and the quantized coefficients are accessed only during image encoding when outputting a code indicating the absolute values of the quantized coefficients when outputting a code denoting the maximum significant digits and outputting a code denoting the signs of the quantized coefficients, which This allows you to more easily encode the image.

In addition, there are cases in which a code indicating the maximum significant digit and a code indicating the signs of the quantized coefficients are 1 bit or 0 bits, so that in the case of encoding such an image, an image can actually be encoded by accessing the quantized coefficients only twice. In addition, in the case of image decoding, access to the quantized coefficients is required to be performed only once, which makes it easier and faster to decode the images.

In the image encoding device 111 and in the image decoding device 211, the quantized subband coefficients are encoded and decoded in raster scan order, and therefore there is no need to place the quantized coefficients in the buffer, thereby reducing the delay caused by encoding and decoding.

In addition, the actual encoding and decoding of the image in YUV 4: 2: 2 format with a size of 1920 pixels horizontally x 1080 pixels vertically when using the operation instruction OKMD (in which w = 4) leads to the following results. It should be noted that during the encoding, the image is subjected to wavelet transform and decomposed into five levels or subbands, and then the quantized coefficients obtained from the quantized wavelet coefficients for each subband are encoded. In addition, the functional blocks necessary for encoding and decoding (for example, the entropy encoding module 123 of Fig. 33 and the entropy decoding module 221 of Fig. 27) and the functional blocks for encoding and decoding an image in JPEG 2000 are implemented by ensuring the execution of the CPU ( with a frequency of 3.0 GHz) type Pentium (registered trademark) 4 (trademark of Intel Corporation) of specified programs.

In the case of encoding one image frame using the conventional JPEG 2000 method, the code size was 291571 bytes and the time required for encoding was 0.26157 seconds. In addition, the time required to decode the encoded image was 0.24718 seconds.

Conversely, in the case of encoding one image frame using the entropy encoding unit 123 shown in FIG. 33, the code size was 343840 bytes and the time required for encoding was 0.03453 seconds. In addition, the time required to decode the encoded image using the entropy decoding unit 221 shown in FIG. 27 was 0.02750 seconds.

Moving images are often displayed at 30 frames per second, which allows real-time image processing only if encoding or decoding can be performed within 0.033 (= 1/30) seconds per frame. When using the JPEG 2000 method, the time required for encoding is 0.26157 seconds, and the time required for decoding is 0.24718 seconds, which makes it difficult to implement real-time processing of images, but in the case of encoding an image using the entropy module 123 coding shown in Fig.33, the time required for coding is 0.03453 seconds, so that the images can be processed almost approximately in real time. In addition, in the case of decoding an image using the entropy decoding unit 221 shown in FIG. 27, the time required for decoding is 0.02750 seconds, which allows efficient processing of images in real time.

In the above description, an example of encoding image data or an example of decoding encoded data has been described, in which image data has been encoded, but not limited to image data, and can be used for cases of encoding audio data or the like, or decoding encoded data in which audio data was encoded, for example. For example, in the case of encoding audio data, a code indicating the maximum significant bits w of setpoints expressed by a code entered as sound data, a code indicating absolute values of numerical values, and a code indicating signs of numerical values are output as encoded audio data.

An additional property of the present embodiment is that with the encoding method described in the present embodiment, lossless quantized coefficients can be encoded. Accordingly, quantization with a large step size for the coefficients of the upper band, so that they correspond to the perception properties of human vision, can significantly improve image quality by the amount of generated code. In addition, by reducing the size of the quantization step, which is used in a certain range of space, it is possible to improve image quality in this range of space.

In addition, using the encoding method described with the present embodiment, a set of significant bits in absolute values is encoded. If we say that the section of significant bits of absolute values is subjected to efficiency coding and transmission, in the case when the significant bit of absolute values is N, an extremely large efficiency table is necessary, having 2 ^∧ (N * W) records (which not only increases the load and processing time in the calculations, but also also requires increasing the storage capacity necessary to maintain the efficiency table). Conversely, when using the encoding method described in the present embodiment, there is no need to use such a large table (which reduces not only the load and processing time for calculations, but also the required storage capacity).

In addition, when using arithmetic coding with a higher compression ratio than the efficiency is provided, even using a compression method with arithmetic coding such as JPEG 2000, for example, it is impossible to significantly improve the compression ratio compared to the case of the coding method described in this embodiment. That is, when using the encoding method described in the present embodiment, encoding processing is performed simply, with a high compression ratio.

When using the encoding method described in the present embodiment, the maximum significant bits of the absolute values w of the sets of coefficients are encoded, as a result, the size of the generated code can be reduced using the fact that the significant bits of adjacent coefficients are similar.

In addition, when using the encoding method described in the present embodiment, differential encoding is performed when encoding the maximum significant bits of the absolute values w of the sets of coefficients, as a result of which the size of the generated code can be reduced using the fact that the significant bits of the neighboring coefficients are also similar at this point.

In addition, entropy encoding processing using entropy decoding module 123 and entropy decoding processing using entropy decoding module 221 in accordance with the sixth embodiment described above can be applied to each of the first to fifth embodiments described above, as a result of which the delay time, power consumption, buffer size required for processing, etc., can be further reduced for all image encoding processing and decoding processing Image IAOD. For example, entropy encoding module 123 may be used as entropy encoding module 15 of the image encoding apparatus 1 shown in FIG. 1 (i.e., entropy encoding module 15 may perform entropy encoding processing in the same manner as entropy encoding module 123). In addition, for example, entropy decoding module 221 can be applied as entropy decoding module 21 in the image decoding apparatus 20 of FIG. 9 (i.e., entropy decoding module 21 can perform entropy decoding processing in the same way as entropy decoding module 221).

That is, when applying the processing of entropy coding in accordance with the sixth embodiment, in each of the above described from the first embodiment to the fifth embodiment, it is possible to reduce the size of the generated code, due to the advantageous use of the property of the order of derivation of coefficients from the wavelet transform module in each of the variants execution (that is, given that the significant bits of successive coefficients are similar). In addition, even in the case of reconfiguration of the coefficients, the wavelet transform module performs the wavelet transform with respect to increments of line blocks, as a result of which the property that significant bits of successive coefficients are similar is not significantly affected, and the size of the generated code during the processing of entropy coding changes slightly.

As described above, the entropy encoding processing in accordance with the sixth embodiment and the properties of the processed coefficient data and its expected benefits are similar to the wavelet transform processing described with reference to the first embodiment to the fifth embodiment described above and have a strong affinity for each other. friend. Accordingly, in the entire image encoding processing, more advantages can be expected when applying the entropy encoding processing in accordance with the present embodiment, when processing the wavelet transform described with reference to the first embodiment to the fifth embodiment described above, as compared to the application other encoding methods.

Next, a seventh embodiment of the present invention will be described. This seventh embodiment is an example of a digital triaxial system in which an image encoding apparatus and an image decoding apparatus in accordance with the above described embodiments are used.

A triax system is a system used in television broadcast stations, production studios, etc. Using such a system, at the time of recording, transmitting from the studio or broadcasting live from a remote place, a single triax cable is used connecting the video camera and the camera control module, or a switch for transmitting multiplexed signals, such as image signals, sound signals, image return signals , timing signals, etc., and also for power supply.

Many conventional triax systems have been installed to transmit the above signals in the form of analog signals. On the other hand, in recent years, all systems have become fully digital, and accordingly, the triax systems used in television broadcast stations have also become digital.

In well-known digital triax systems, digital video signals transmitted over a triax cable have so far been uncompressed video signals. The reason for this is that the specifications required with respect to signal delay time are extremely strict in television broadcasting stations, for example, for example, it is required that the delay time from the moment of recording to output to the monitor screen be within the same field (16, 67 ms). Compression encoding systems, such as MPEG2 and MPEG4, which implement a high compression ratio and high image quality, are not used in triax systems, since time equivalent to several frames is required for compression and encoding of a video signal, and decoding of compressed video signals, which means too much a large delay in time.

The methods for encoding and decoding an image in accordance with the present invention described above have an extremely short delay time for inputting image data to obtain an output image within the time of one field, that is, from several lines to several tens of lines, and accordingly can be used in a digital triaxial system.

On Fig illustrates the configuration of an example of a digital triaxial system applicable in the method of encoding and decoding an image in accordance with the present invention. The transmission module 500 and the camera control module 502 are connected via a triax cable (coaxial three-wire cable) 501. Digital video signals and digital audio signals (hereinafter referred to as “main line signals”) from the transmission module 500 to the camera control module 502, which are actually transmitted in broadcast mode transmissions, or use both the content and the audio signals of the intercom and the inverse digital video signals from the camera control module 502 of the video camera module 503 are transmitted via a triaxial cable 501.

Transmission module 500 is integrated, for example, in a camcorder device not shown. Of course, other arrangements can be made, such as a transfer module 500 connected to the camcorder device as an external device for the camcorder device. The camera control module 502 may be a device commonly referred to as, for example, a CCM (CCU, camera control module).

Digital audio signals constitute a small load in the sense of the present invention, therefore, their description will not be given here to simplify the description.

The camera module 503 is made, for example, inside a video camera device not shown in the drawing, and shoots using an unshown image tracking device, such as a CCD (CCD, charge-coupled device), light from an object that is captured through an optical system 550 including lens, focusing mechanism, optical zoom mechanism, aperture adjustment mechanism, etc. The image pickup apparatus converts the received light into electrical signals using photoelectric conversion, and furthermore performs predetermined signal processing for outputting digital video signals of the main range. These digital video signals are displayed, for example, in the PID-BP format (HD-SDI, high resolution serial data interface) and output.

Also, a display module 551, used as a monitor, and an intercom 552, used to exchange sound with external devices, are connected to the camera module 503.

The transmission module 500 has a video encoding module 510, a video decoding module 511, a digital modulation module 512 and a digital demodulation module 513, amplifiers 514 and 515, and a video separation / synthesis module 516.

Digital video signals of the main range, displayed, for example, in the PID-BP format, are transmitted from the camera module 503 to the transmission module 500. Digital video signals are compressed and encoded in the video encoding module 510 so that they become a code stream that is transmitted to the digital modulation module 512. Digital modulation module 512 modulates the transmitted code stream into signals in a format suitable for transmitting data via triax cable 501 and output them. The signals output from the digital modulation module 512 are transmitted to the video separation / synthesis module 516 through an amplifier 514. The video separation / synthesis module 516 transmits the transmitted signals to the triax cable 501. These signals are received in the camera control module 502 via the triax cable 501.

The signals output from the camera control module 502 are received in the transmission module 500 via a triaxial cable 501. The received signals are transmitted to the separation / synthesis module 516, where a part of the digital video signals and a part of other signals are separated. Of the received signals, part of the digital video signals is transmitted through an amplifier 515 to a digital demodulation module 513, and the signals modulated into signals in a format suitable for transmission via a triax cable 501 are demodulated on the side of the camera control module 502, and the code stream is restored.

The code stream is transmitted to the video decoding module 511, compression encoding is decoded, and digital video signals in the main range are obtained. The decoded digital video signals are displayed in PID-BP format and output and transmitted to the camera module 503 as return digital video signals. Return digital video signals are transmitted to a display module 551 connected to a video camera module 503 and used for tracking by a camera operator.

The camera control module 502 has a video separation / synthesis module 520, amplifiers 521 and 522, a preprocessing module 523, a digital demodulation module 524 and a digital modulation module 525, and a video signal decoding module 526, as well as a video encoding module 527.

The signals output from the transmission module 500 are received in the camera control module 502 via a triaxial cable 501. The received signals are transmitted to the video separation / synthesis module 520. The video separation / synthesis module 520 transmits the signals transmitted therein to the digital demodulation module 524 through an amplifier 521 and a preprocessing module 523. It should be noted that the pre-processing module 523 has a gain control module for adjusting the gain of the input signals, a filter module for performing a predetermined filtering of the input signals, etc.

Digital demodulation module 524 demodulates the signals modulated into signals in a format suitable for transmission over the triax cable 501 on the side of the transmission module 500, and restores the code stream. The code stream is transmitted to the video decoding module 526, where compression encoding is decoded to receive digital video signals in the main range. The decoded digital video signals are displayed in PID-BP format and output, and transmitted to an external device, as the main linear signals.

The returned digital video signals and digital audio signals are used externally in the camera control unit 502. Digital audio signals are transmitted through an intercom 552 camera operator, for example, for use in transmitting external audio instructions to the camera operator.

The returned digital video signals are subjected to encoding 527 of the video signal, decoded with compression and transmitted to the digital modulation module 525. Digital modulation module 525 modulates the transmitted code stream into signals in a format suitable for transmission over triax cable 501, and outputs them. The signals output from the digital modulation module 525 are transmitted to the video separation / synthesis module 520 through a preprocessing module 523 and an amplifier 522. The video separation / synthesis module 520 multiplexes these signals with other signals and transmits them via a triax cable 501. The signals are received in the module 503 camcorders via triax 501.

In such a seventh embodiment of the present invention, the image encoding apparatus and the image decoding apparatus described in the above embodiments are respectively applied to the video encoding module 510 and the video encoding module 527, and the video decoding module 511, and the video decoding module 526 .

In particular, the second embodiment of the present invention is arranged such that the processing of various elements in the image encoding device and the image decoding device can be performed in parallel, which can significantly reduce the time delay when outputting images captured by the camera module 503 from the camera control unit 502, and the delay in the return of digital video signals transmitted from external devices and transmitted from the camera control module 502 to the camera module 503 and, accordingly, yaemyh in the seventh embodiment of the present invention.

In addition, in the case of the system shown in FIG. 37, the signal processing capabilities and the memory capacity can be set appropriately in each of the transmission module 500 and the camera control module 502, so that the coefficient data reconfiguration processing will be performed either on the module side 500 of the transmission, either on the side of the camera control module 502, and entropy encoding can be performed in the same way, either before or after reconfiguration processing.

That is, on the side of the transmission module 500, the video encoding module 510 performs wavelet transform and entropy encoding in accordance with the method of the present invention for digital video signals transmitted thereto, and outputs a code stream. As described above, after entering a certain number of lines corresponding to the number of taps of the filter used for the wavelet transform, and in accordance with the number of input levels of the separation of the wavelet transform, the video encoding module 510 starts the wavelet transform. In addition, as described above with reference to FIG. 5, FIG. 6, FIG. 11, etc., after accumulating coefficient data necessary for the components in the image encoding device and in the image decoding device, the processing is sequentially performed by these components . After processing on the bottom line of one frame or one field, processing of the next one frame or one field begins.

This is also true for digital video signals returned by transmission from the side of the camera control unit 502 to the side 500 of the transfer unit. That is, on the side of the camera control unit 502, the video signal encoding unit 527 performs wavelet transform and entropy encoding in accordance with the present invention on the returned digital video signals transmitted from external devices, and a code stream is output.

This often leads to the case where it is permitted that the returned digital video signals have lower image quality than the digital video signal for the main signal line. In this case, the bit rate during encoding in the video encoding unit 527 may be reduced. For example, the video encoding unit 527 controls the bit rate control unit 14 so that the bit rate of the entropy encoding processing in the entropy encoding unit 15 is lower. In addition, one can consider such an arrangement in which, for example, on the side of the camera control module 502, the conversion processing is performed at a higher level of separation with the wavelet transform module 10 to the video encoding module 527, and on the side of the transmission module 500, the inverse wavelet transform in the module The 23 inverse wavelet transforms on the side of the video encoding module 511 are set to a lower separation level. The processing in the video encoding unit 527 by the camera control unit 502 is not limited to this example, and various other types of processing can be considered, such as supporting a low level of separation for the wavelet transform so as to reduce the load on the conversion processing.

Next, an eighth embodiment of the present invention will be described. In this eighth embodiment of the present invention, the transmission of encoded data encoded in the image encoding apparatus in accordance with the present invention to the side of the image decoding apparatus is performed using a wireless data transmission channel. On Fig presents a configuration example of a wireless data transmission system in accordance with the eighth embodiment of the present invention. It should be noted that in the example shown in FIG. 38, video signals are transmitted in one direction from the side of the video camera or transmission module 600 (hereinafter abbreviated as “transmission module 600”) to the receiving device 601. Bidirectional data transmission between the data transmission unit 600 and the receiving device 601 may be performed for video signals and other signals.

A transmission module 600 is integrated in a video camera device not shown, having, for example, a video camera module 602. Of course, other arrangements can be made, such as a transmission module 600 connected to the video camera device, as an external device for the video camera device having the video camera module 602.

The video camera module 602 has a specific optical system, an image pickup device, such as, for example, a CCD, and a signal processing module for outputting signals output from the image pickup device in the form of digital video signals. Such digital video signals are displayed, for example, in the PID-BP format and output, for example, from the camera module 602. Of course, the digital video signals output from the camera module 602 are not limited to this example, and may also have other formats.

The transmission module 600 has a video coding module 610, a digital modulation module 611, and a wireless module block 612. In transmission module 600, digital baseband video signals are mapped, for example, to PID-BP format and output from video camera module 602. Digital video signals are subjected to compression encoding using wavelet transform and entropy encoding, in accordance with the compression encoding method in accordance with the present invention, in video encoding module 610 so that they become a code stream that is transmitted to digital modulation module 611. Digital modulation module 611 digitally modulates the transmitted code stream into signals in a format suitable for wireless data transmission and outputs them.

In addition, digital audio signals and other signals, such as, for example, given commands and data, are also transmitted to digital modulation module 611. For example, the camera module 602 has a microphone, with which the captured sounds are converted into audio signals, and in addition, the audio signals are A / D converted and output as digital audio signals. In addition, the camera module 602 may output certain commands and data. Commands and data may be generated in the camera module 602, or an operation module may be provided in the camera module 602 so that commands and data will be generated in accordance with user operations performed in the operations module. In addition, an arrangement may be made in which an input device for inputting a command and data will be connected to the camera module 602.

Digital modulation module 611 digitally modulates and outputs these digital audio signals and other signals. The digital modulated signals output from the digital modulation module 611 are transmitted to the wireless module unit 612 and transmitted wirelessly through the antenna 613 in the form of radio waves.

After receiving the AZP (ARQ, automatic repeat request) from the receiving device 601, the wireless module unit 612 transmits a notification in the form of such an AZP to the digital modulation module 611 to request data retransmission.

The radio waves transmitted through the antenna 613 are received by the antenna 620 on the side of the receiving device 601, and they are transmitted to the wireless module unit 621. The wireless module unit 621 transmits digital modulated signals based on the received radio waves to the preprocessing module 622. The preprocessing unit 622 performs predetermined signal processing, such as, for example, gain control with respect to the transmitted digital modulated signals, and transmits them to the digital demodulation unit 623. Digital demodulation module 623 demodulates the transmitted digital modulated signals and restores the code stream.

The code stream restored in digital demodulation module 623 is transmitted to a video decoding module 624, where decoding is performed for compression encoding using the decoding method in accordance with the present invention, and digital baseband video signals are obtained. The decoded digital video signals are displayed, for example, in PID-BP format and output.

Digital audio signals and other digitally modulated signals on the side of transmission module 600 are also transmitted to digital demodulation module 623 and are transmitted. Digital demodulation module 623 demodulates the signals in which these digital audio signals and other signals were digitally modulated, and restores and outputs digital audio signals and other signals.

In addition, the preprocessing unit 622 performs error detection in accordance with a predetermined method with respect to the received signals transmitted from the wireless unit 621, and in the case where an error is detected, such as, for example, receiving a frame with an error, outputs the ALR. This AZP is transmitted to a wireless module unit 621 and transmitted through an antenna 620.

With this configuration, the transmission module 600 is integrated into the relatively small size video camera device, having, for example, the video camera module 602, the monitor device is connected to the reception device 601, and the digital video signals output from the video decoding module 624 are transmitted to the monitor device. When the reception device 601 is within the transmission distance of the radio waves transmitted from the wireless module unit 612 from the video camera device having the built-in transmission module 600, images captured by the video camera device can be viewed on the monitor device with a low delay, for example, with a delay within one field or one frame.

It should be noted that in the example shown in Fig. 38, data transmission between the transmission module 600 and the reception module 601 will be performed using wireless data transmission, for example, in the form of video signals transmitted over a wireless data transmission channel, but this arrangement is not limited to this example . For example, the transmission module 600 and the receiving module 601 may be connected through a network such as the Internet. In this case, each block 612 of the wireless module on the side of the transmission module 600 and the block 621 of the wireless module on the side of the reception module 601 are data transmission interfaces that allow data transmission using a PI (IP, Internet protocol).

Various applications may be considered for the system in accordance with this eighth embodiment. For example, a system in accordance with such an eighth embodiment may be applied to a video conferencing system. This layout example uses a simple camcorder device that can be connected via USB (USB, universal serial bus) to a computer device, such as a personal computer, with a video encoding module 610 and a video decoding module 624 on the side of the computer device. The video encoding module 610 and the video decoding module 624 installed in the computer device may have a hardware configuration, or may be implemented as software running in a computer device.

For example, each of the participants in a video conference can be equipped with a computer device and a video camera device that connects to a computer device, and the computer device is connected to a server device to provide video conferencing system services over a cable network or wireless network. The video signals output from the camcorder device are transmitted to the computer device via the USB cable, and encoding processing, in accordance with the present invention, is performed in the video signal encoding module 610 in the computer device. A computer device transmits a code stream in which video signals have been encoded to a server device or the like via a network.

The server device transmits the received code stream to the computer device of each of the participants through the network. The code stream is received at the computer device of each of the participants and is subjected to decoding processing, in accordance with the present invention, in the video signal decoding unit 624, in the computer device. Image data output from the video decoding unit 624 is displayed on the display unit of the computer device as an image.

That is, the video signals captured by the video camera devices of other participants are displayed in the display modules of the computer devices of each of the participants. Accordingly, in the eighth embodiment of the present invention, the delay time from the moment of encoding the video signals captured by the video camera device to their decoding in the computer device of other participants will be short, which reduces the feeling of unnatural images of other participants displayed in the display modules of computer devices displayed participants.

In addition, an arrangement may be provided in which the video encoding module 610 is mounted on the side of the camcorder device. For example, a transmission module 600 may be integrated in a camcorder device. This configuration eliminates the need to connect the camcorder device to another device, such as a computer device or the like.

Such a system, consisting of a video camera device with an integrated transmission module 600 and a receiving device 601, can be applied in various fields, in addition to the video conferencing system described above. For example, as shown schematically in FIG. 39, such a system can be used in a home gaming console. 39, a transmission module 600 in accordance with an eighth embodiment of the present invention is integrated in the camcorder device 700.

In the main module 701 of the home game console, for example, via a bus, a CPU, RAM, ROM are connected, a CD-ROM compatible disc drive (CD-ROM) and a DVD-ROM (digital universal ROM) are connected disk), a graphical control module for converting the display control signals generated by the CPU into video signals and output, and a sound reproducing module for reproducing audio signals, etc., that is, having a configuration generally similar to th configuration of the computing device. The main module 701 of the home game console is fully controlled by the CPU, in accordance with programs previously stored in the ROM, or programs recorded on a CD-ROM or DVD-ROM installed in the disk drive device. RAM is used as working memory for the CPU. A reception device 601 is integrated in the main module 701 of the home gaming console. Digital video signals output from the reception device 601 and other signals are transmitted to the CPU, for example, via a bus.

We can say that in such a system, for example, the main module 701 of the home game console, the work of the game software is performed, which can take images in the form of digital video signals transmitted from an external source, like the images used in the game. For example, such game software allows the use of images in the form of digital video signals transmitted from an external source, such as images used in the game, and also recognizes the movements of people (players) in this image and performs operations corresponding to the recognized movements.

The video camera device 700 encodes the captured digital video signals using the encoding method, in accordance with the present invention, in the video encoding module 610, in the built-in transmission module 600, modulates the code stream in the digital modulation module 611 and transmits to the wireless module 612 for transmission through the antenna 613. The transmitted radio waves are received using the antenna 620 of the receiving device 601, which is integrated in the main module 701 of the home game console, the received signals are transmitted to the digital demodulation module 623 through the block 621 wireless modules and pre-processing module 622. The received signals are demodulated in a digital demodulation module 623 into a code stream and transmitted to a video signal decoding module 624. The video decoding unit 624 decodes the transmitted code stream using the decoding method in accordance with the present invention and outputs digital video signals in the base band.

Digital video signals in the main band output from the video decoding module 624 are transmitted via a bus in the main module 701 of the home game console and temporarily stored, for example, in RAM. After reading the digital video signals stored in RAM, in accordance with a predetermined program, the CPU can detect the movement of people in the image provided as digital video signals, and use this image in the game.

Due to the short delay time from the moment of taking images using the video camera device 700 to receiving digital video signals encoded as a code stream that is decoded in the main module 701 of the home game console and receiving images therein, the response of the gaming software operating in the main module 701 is improved the home console of the business game with respect to player movement, thereby improving the response of the home game console to player movement, thereby improving overall functionality s games.

It should be noted that such a camcorder device 700 used in a home gaming console often has a simple configuration due to price, size, etc. limitations, and it should be assumed that a CPU with large processing power and a large capacity storage device , such as in a computer device, may not be used.

That is, typically, the video camera device 700 is a peripheral device of the home game console main module 701, which is only needed for playing games using the video camera device 700, and is not the device necessary for playing games in the main home game console module 701. In this case, the camcorder device 700 is often sold separately from the main module 701 of the home gaming console (the so-called optional, separately sold device). In this case, the installation of a high-performance CPU and a high-capacity storage device in the camcorder device 700, which are sold at a high price, can usually lead to a decrease in the sales volume of the product. Thus, this can lead to a decrease in sales of games that use the camcorder device 700, which can lead to a decrease in revenue. In addition, in home games, in particular, the proportion of ownership often greatly affects the amount of goods sold, so that a low proportion of ownership of the camcorder device 700 can lead to an additional reduction in the amount of goods sold.

On the other hand, when selling a large number of low-cost video camera devices 700, to increase the ownership ratio, you can improve the sale of home games that use the video camera device 700 and increase their popularity, and this, in turn, can stimulate additional motivation to purchasing the main module 701 home gaming console. Given this perspective, the camcorder device 700 often preferably also has a simple configuration.

In this case, it is possible to provide an arrangement in which the wavelet transform is performed in the video encoding module 610, the transform module 600, integrated in the camera device 700, at a low separation level. This reduces the capacity requirement of the storage device used with the coefficient reconfiguration buffer module.

In addition, you can consider the layout that uses the configuration of the image encoding device, an example of which is shown in Fig.12, which was described in the third embodiment, for the module 610 encoding a video signal. In addition, the application of the configuration of the image encoding device as shown in FIG. 15, which was described in the fourth embodiment, in the video encoding module 610, eliminates the need to perform reconfiguration processing of the wavelet transform coefficient data on the side of the video encoding module 610, as a result which can further reduce the load on the side of the camcorder device 700, which is preferred. In this case, there is a need to use an image decoding apparatus, an example of which is shown in FIG. 16, which was described in the fourth embodiment, as a video decoding module 624 in a reception device 601, integrated on the side of the main module 701 of a home game console.

It should be noted that the camcorder device 700 and the home game console main module 701 have been described above as devices connected using a wireless data channel, but this arrangement is not limited to such an example. That is, the camcorder device 700 and the home game console main module 701 can be connected via cable through interfaces such as USB, IEEE 1394, or the like.

As described above, another significant advantage of the present invention is that it can be applied in various forms, and can be easily used in a wide variety of uses (that is, it has a high degree of versatility).

The sequences of the processing described above can be implemented using hardware or using software. In the case of the implementation of the processing sequence in the form of software, the programs comprising the software are installed in a computer that has built-in specialized hardware, or they can be installed in a public computer, or in an information processing device of an information processing system composed of many devices, which allow you to perform various functions using different types of programs installed on them from the recording medium of the program.

40 is a block diagram illustrating an example configuration of an information processing apparatus for executing the above-described processing sequences using a program.

As shown in FIG. 40, the information processing system 800 is a system configured as an information processing device 801, a storage device 803 connected to the information processing device 801 using a PCI bus 802, NPS (VTR, tape recorder) 804- 1 - UZL 804-S, which are a lot of video recorders (UZL) and a mouse 805, a keyboard 806, and an operation controller 807, designed to perform operations entered by the user of these devices, and is a system that performs code processing tion image and image decoding processing, and the like, as described above, using an installed program.

The information processing device 801 of the information processing system 800, for example, can encode the contents of the moving image stored in the large capacity storage device 803, which is configured as an MNDI (RAID, redundant array of independent disk drives) and stores the obtained encoded data in the storage device 803, decoding the encoded data stored in the storage device 803 and storing the obtained decoded image data (moving image content) in the storage device 803 It records encoded data or decoded image data on a video tape using UZL 804-1 - UZL 804-S, etc. In addition, the information processing device 801 is configured to capture the contents of a moving image recorded on video tapes installed in the UZL 804-1 - UZL 804-S in the storage device 803. In this case, an arrangement may be provided in which the information processing device 801 encodes the contents of a moving image.

The information processing module 801 has a microprocessor 901, MGO (GPU, graphic processing module) 902, RAM-FWM (XDR-RAM, RAM with an extremely high data transfer rate) 903, south bridge 904, HDD 905, interface (IF, I / F ) 906 USB, and the codec 907 sound input / output.

MGO 902 is connected to microprocessor 901 through a specialized bus 911, RAM-PMC 903 is connected to microprocessor 901 through a specialized bus 912. The south bridge 904 is connected to the input / output controller 944 of microprocessor 901 through a specialized bus. Also, the HDD 905, the USB interface 906 and the sound input / output codec 907 are connected to the south bridge 904. Loudspeaker 921 is connected to sound input / output codec 907. In addition, the display 922 is connected to the IGO 902.

Also connected to the south bridge 904 are a mouse 805, a keyboard 806, an UZL 804-1-804-S, a storage device 803, and an operation controller 807, via a PCI 802 bus.

A mouse 805 and a keyboard 806 accepts user input and transmits signals to the microprocessor 901 indicating the content of user input via PCI bus 802 and south bridge 904. Storage device 803-1 and UZL 804-1 - UZL 804-S can record play set data.

The drive 808 can optionally be connected to a PCI bus 802, in which removable media 811 is mounted, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory device, etc., respectively, and a computer program read from it is installed in accordance with the need for HDD 905.

The microprocessor 901 has a multi-core configuration in which the core 941 of the main general-purpose CPU, designed to run basic programs such as OS (OS, operating system) or the like, the core 942-1 of the secondary CPU is the core 942-8 of the secondary CPU, which represent there are many (eight in this case) KOHI-type signal processors (RISC, a computer with a limited set of instructions) connected to the core 941 of the main CPU via an internal bus 945, a memory controller 943 designed for To control the memory device, RAM-903, having a capacity of, for example, 256 [MB], and a controller 944 I / O (I / O, input / output), designed to control the input and output of data together with the south bridge 904, integrated on a single chip, with the implementation of the operating frequency, for example, 4 [GHz].

During startup, the microprocessor 901 reads the necessary application programs stored in the HDD 905 based on the control program stored in the HDD 905, and transfers them to the RAM-PMC 903, and then performs the necessary control processing based on the application programs and operations performed by the operator .

In addition, when executing the software, the microprocessor 901 can implement image encoding processing and image decoding processing, in accordance with the above-described embodiments, by transmitting the code streams obtained as a result of encoding to the HDD 905, for their storage, through the south bridge 904, can perform the transfer of data, which are, for example, images for playback, the contents of a moving image obtained by decoding in the IHO 902, for display on the display 922, etc. d.

Although approaches to using the CPU cores of microprocessor 901 are optional, an arrangement may be provided in which, for example, the core CPU core 941 performs processing related to the control of image encoding processing and image decoding processing, and eight auxiliary cores 942-1-942-8 CPUs perform various types of processing, such as wavelet transform, coefficient reconfiguration, entropy coding, entropy decoding, inverse wavelet transform, quantization, inverse quantization, etc., simultaneously and in parallel, as described, for example, with reference to FIG. 11. In this case, an arrangement can be obtained in which the core 941 of the main CPU assigns processing to each of the eight cores 942-1-942-8 of the auxiliary CPUs in increments in the form of string blocks, thereby performing image encoding processing and image decoding processing, simultaneously in parallel, incrementally in blocks of rows, in the same way as in the case described with reference to FIG. That is, the efficiency of image encoding processing and image decoding processing can be improved, the delay time of all processing can be reduced, and in addition, the load, processing time and memory capacity necessary for processing can be reduced. Of course, each processing can also be performed using other methods.

For example, an arrangement may be provided in which a part of eight cores 942-1 - cores 942-8 of the auxiliary microprocessor 901 perform encoding processing, and the others simultaneously and simultaneously execute decoding processing.

In addition, for example, in the case where an independent encoder or decoder or codec processing device is connected to the PCI PCI bus, eight cores 942-1-942-8 of the auxiliary CPU of microprocessor 901 can control the processing performed by these devices through the south bridge 904 and an 802 PCI bus. In addition, in the case where a plurality of such devices are connected, or in the case when these devices include a plurality of decoders or encoders, eight cores 942-1-942-8 of an auxiliary CPU of the microprocessor 901 can control such that a plurality of decoders or coders will jointly share processing.

In this case, the main core 941 of the CPU controls the operation of eight cores 942-1-942-8 of the auxiliary CPU, assigning processing to each core of the CPU, receiving processing results, etc. In addition, the main CPU core 941 also performs other processing than that which these auxiliary CPU cores perform. For example, the main CPU core 941 receives commands transmitted through a mouse 805, a keyboard 806, or an operation controller 807 via a south bridge 904 and performs various types of processing corresponding to these commands.

The IGO 902 performs the final data processing for texture mapping, and so on, to reproduce the contents of the game image or the content of the moving image to be displayed on the display 922, and also controls the functions for calculating the coordinate transformation when displaying multiple game images for the content of the moving images and still image images on the display 922, immediately, processing the increase / decrease of game images used as the moving image contents and still images of still image contents, etc., whereby it is possible to reduce the burden of the microprocessor 901.

MGO 902, under the control of microprocessor 901, performs specified signal processing in relation to the image data supplied to it, which is the content of the moving image, and image data, which is the content of the still image, transmits image data and display data obtained as a result of this processing to a display 922, and displays image signals on a display 922.

Now, game images, which are a lot of contents of a moving image, decoded simultaneously and in parallel using eight cores 942-1-942-8 of the auxiliary CPU of microprocessor 901, are subjected to data transmission to MGO 902 via bus 911, while the transmission speed is, for example, up to up to 30 [GB / s], so even complex images with special effects can be continuously displayed at high speed.

In addition, the microprocessor 901 performs sound bias processing for the sound data corresponding to the image data and the sound data corresponding to the contents of the moving image, and transmits the edited sound data obtained as a result of this processing to the loudspeaker 921 through the south bridge 904, and the input code 907 audio output, for outputting sound based on audio signals through a speaker 921.

In the case of implementing the above-described processing sequences using software, the program constituting the software is installed via a network or a recording medium.

Such a recording medium includes not only removable media 811 shown in FIG. 40, such as magnetic disks (including floppy disks), optical disks (including CD-ROMs and DVDs), magneto-optical disks (including MD), a semiconductor memory device, and etc., installed separately from the main device module for distributing programs to users, but also HDD 905, a storage device 803, in which programs, etc., delivered to the user assembled with the main module of the device, are recorded. Of course, the recording medium may also be a semiconductor memory device, such as a ROM or a flash memory device or the like.

In the above description, it is indicated that the microprocessor 901 is made up of eight auxiliary CPU cores, but the invention is not limited thereto, and the number of auxiliary CPU cores is arbitrary. In addition, with regard to microprocessor 901, an arrangement may be used that uses a CPU configured as a single-core (single core) CPU instead of the main CPU core and auxiliary CPU cores. In addition, a plurality of CPUs can be used in place of the microprocessor 901, a plurality of information processing devices can be used (that is, a program for executing processing in accordance with the present invention is executed on a plurality of devices working together).

Although the steps describing the program recorded on the program recording medium in the present description can, of course, be performed in the described time sequence, they are not limited to this time sequence and can be performed in parallel, or separately.

In addition, the system used in the present description relates to all equipment, which is made of many devices (devices).

It should be noted that in the above description, the configurations presented as a separate device can be divided so that they will consist of many devices. Conversely, in the above description, configurations described as being made of a plurality of devices can be combined and can be made as a single device. In addition, other configurations may be added to the configuration of these devices beyond those described above. In addition, part of the configuration of one device may be included in the configuration of another device, provided that the configuration and operation of the entire system essentially remains the same.

Industrial applicability

The present invention described above can preferably be applied to various devices or systems as long as they are intended for encoding with compression and transmission of images, and the compression code is decoded and output at the destination of transmission. The present invention, in particular, is mainly used with devices or systems in which a small delay is required from the moment of encoding with compression to the moment of decoding and outputting images.

For example, the present invention, preferably, can be used in remote medical diagnostics and in the treatment using manipulators, when viewing images taken, for example, with a video camera. In addition, the present invention can preferably be used for compression coding and transmission of digital video signals, and decoding of digital video signals subjected to compression coding in systems such as those used in broadcast stations and the like.

In addition, the present invention can be applied in systems designed for the distribution of live video, in educational systems in which students and teachers can interact interactively with each other, etc.

In addition, the present invention can be applied to transmit image data captured by mobile terminals having image pickup functions, such as cell phones with camera functions, in video conferencing systems, in surveillance systems, for recording images captured by security cameras using a recording device and etc.

Claims (22)

1. An encoding device for encoding image data, comprising filter means for processing filtering said image data for each block of rows, including image data of a plurality of rows necessary to generate coefficient data corresponding to one row of a subband of at least a component of the lowest band, and generating a plurality of subbands consisting of coefficient data distributed over a frequency range; storage means for storing coefficient data generated by said filter means for each row block; and coefficient reconfiguration means for reconfiguring the specified coefficient data stored in the indicated storage means for each row block in the order in which the synthesis processing should be performed to generate image data by synthesizing coefficient data of a plurality of subbands that have been divided into frequency ranges, wherein said coefficient reconfiguration means reconfigures said coefficient data in order from the components of the lower floor sy to highband components for each line block; encoding means for encoding coefficient data reconfigured by said coefficient reconfiguration means; and control means for controlling said filter means, said storage means, said coefficient reconfiguration means and said coding means for parallel operation for each block of rows. 2. The encoding device according to claim 1, wherein said filter means, said save means and said coefficient reconfiguration means perform each process in parallel. 3. The encoding device according to claim 1, wherein said filter means performs said filtering processing in a vertical direction and in a horizontal direction on a screen corresponding to said image data. 4. The encoding device according to claim 1, in which at least one of the many conclusions and the number of resolution levels of the specified filtering processing performed by the specified filter means are determined in accordance with the target delay time. 5. The encoding device according to claim 1, wherein said filter means performs wavelet filtering processing, wherein the subband coefficient data of the lower band component obtained by said filtering processing is further subjected to said filtering processing. 6. The encoding device according to claim 5, wherein said filter means performs said wavelet filtering processing using a lifting technique. 7. The encoding device according to claim 5, in which, when the specified filter means performs the specified filtering processing at the level = X + 1 of the resolution using the specified lifting technique, it is performed for coefficient data calculated as a subband of the lower band component with filtering processing at the level = X resolution. 8. The encoding device according to claim 5, wherein said storage means further comprises first buffer means for containing subband coefficient data of a lower band component generated during said wavelet filtering processing performed by said filter means; and second buffer means for containing subband coefficient data of the upper band component generated during said wavelet filtering processing performed by said filter means. 9. The coding apparatus of claim 8, wherein said means of the second buffer comprises sub-band coefficient data of components of another band other than said lowest band, while coefficient data for the lowest band of the band component is generated by said filter means. 10. The encoding device according to claim 1, wherein as soon as said coefficient data is reconfigured by said coefficient reconfiguration means, said encoding means sequentially performs entropy encoding of the reconfigured coefficient data in reconfiguration order. 11. The encoding device according to claim 1, wherein said encoding means performs encoding of the packet on a plurality of rows of coefficient data within one subband. 12. The encoding device according to claim 1, wherein said encoding means performs encoding on the rows of all subbands constituting the row block with respect to the data of the coefficient stream aligned in the one-dimensional direction in the order from the lower band to the upper band. 13. The encoding device according to claim 1, further comprising a packetizing means for adding a predetermined header to the encoded data generated by said encoding means and packetizing the header and data body; transmission means for transmitting said packet generated by said packetizing means; and control means for controlling said filter means, said saving means, said coefficient reconfiguration means, said encoding means, said packetizing means, and said transmission means for parallel operation for each row of the block. 14. The encoding device according to claim 1, further comprising a packetizing means for adding a predetermined header to the encoded data generated by said encoding means and packetizing the header and data body; and transmission means for transmitting said packet generated by said packetizing means; wherein said filtering means, said saving means, said coefficient reconfiguration means, said encoding means, said packing means and said transmission means work in parallel for each row block. 15. The encoding device of claim 14, wherein said header includes identification information for identifying said block of a line on a screen, data length of a data body and encoding information. 16. The encoding device according to clause 15, wherein said encoding means comprises a quantization means for quantizing said coefficient data generated by said filter means; source encoding means for encoding a source based on quantization result coefficients obtained from said coefficient data quantized by said quantization means; in which the specified encoding information includes information about the size of the quantization step performed by the specified quantization means. 17. An encoding method for encoding image data, comprising a filtering step for performing filtering processing of said image data for each block of rows including image data of a plurality of rows necessary to generate coefficient data corresponding to one row of a subband of at least at least a component of the lowest band, and generating a plurality of subbands consisting of coefficient data distributed over a frequency band; a storage step for storing coefficient data generated at a specified filtering step for each block of rows; a coefficient reconfiguration step for reconfiguring the specified coefficient data stored in the indicated storage step for each row block in the order in which the synthesis processing is to be performed to generate image data by synthesizing coefficient data of a plurality of subbands that have been divided into frequency bands wherein the reconfiguration of said coefficient data is carried out in order from the components of the lower band to the components of the upper band, for each Lok rows; a coding step for coding coefficient data reconfigured at said reconfiguration step; however, all the above steps are carried out in parallel for each block of rows. 18. A decoding device for decoding encoded data in which image data has been encoded, comprising decoding means for decoding said encoded data and generating coefficient data included in a plurality of subbands; storage means for storing coefficient data generated by said decoding means for each block of lines including image data of a plurality of lines necessary to generate coefficient data corresponding to one line of a subband of at least a component of the lowest band; coefficient reconfiguration means for reconfiguring the specified coefficient data stored in the specified storage means for each row block in the order of synthesizing the coefficient data of a plurality of subbands that have been divided into frequency bands, wherein the reconfiguration of the coefficient data is carried out in order from the lower band components to the components of the upper strip, for each block of rows; filter means for filtering processing for said coefficient data reconfigured by said coefficient reconfiguration means and synthesizing coefficient data of a plurality of subbands distributed over a frequency band to generate said image data; and control means for controlling said decoding means, said saving means, said coefficient reconfiguration means and said filter means for parallel operation in each block of rows. 19. The decoding apparatus of claim 18, wherein said decoding means, said saving means, said coefficient reconfiguration means, and said filter means each performs processing in parallel. 20. The decoding device according to claim 18, wherein said filter means performs said filtering processing using a lifting technique. 21. The decoding apparatus according to claim 18, wherein said decoding means decodes encoded data in which lines of all subbands comprise said block of lines that has been encoded and aligned in a one-dimensional direction. 22. A decoding method for decoding encoded data in which image data has been encoded, comprising a decoding step for decoding said encoded data and generating coefficient data included in a plurality of subbands; a storage step for storing coefficient data generated by said decoding means for each block of strings including image data of a plurality of rows necessary to generate coefficient data corresponding to one row of a subband of at least a component of the lowest band; a coefficient reconfiguration step for reconfiguring the specified coefficient data stored at the indicated storage step for each row block in the order of synthesizing the coefficient data of a plurality of subbands that have been divided into frequency bands, wherein the reconfiguration of the coefficient data is carried out in an order from the lower band components to top bar components for each row block; a filtering step for processing filtering said coefficient data reconfigured at said coefficient reconfiguration step and synthesizing coefficient data of a plurality of subbands distributed over a frequency band to generate said image data, all of the above steps being carried out in parallel for each block of rows.

Images