KR100233764B1 - Video signal coder - Google Patents

Abstract

Frame / field motion detection circuits 22 and 21 for detecting the motion vector and the absolute difference between each pixel between frames and odd and even fields, and a motion prediction mode for selecting an efficient mode in motion compensation mode on a frame / field basis. Frame memory on the basis of the output of the circuit 25, the selector 24, the block mode decision circuit 25 for selecting a mode with high efficiency in each block mode of orthogonal number of frames / fields, and the circuit 25; An address generator 11 for controlling the group and a frame memory root 20 with a motion compensator operating on the basis of the outputs of the circuits 23 and 25 are provided.

For moving pictures having a field structure, even if the image is small in motion, the image is high in motion, or both are common, field processing or frame processing can be made efficient.

Description

Video signal coder

1 is a block diagram showing a schematic configuration of a high efficiency encoding device of an image signal of the first embodiment.

2 is a block diagram showing a schematic configuration of a high efficiency encoding device of an image signal according to the second embodiment.

3 shows a macro block.

4 is a diagram showing a macroblock in a frame processing mode.

5 is a diagram showing a macroblock in the field processing mode.

6 is a diagram for explaining a state of an encoding process of the embodiment apparatus.

FIG. 7 shows a unit block of DCT processing in the frame processing mode / field processing mode in the format of one digital VTR. FIG.

FIG. 8 is a diagram showing a state of motion prediction in the example of FIG.

9 is a diagram showing an example other than FIG.

FIG. 10 is a diagram showing a state of motion prediction in the example of FIG.

FIG. 11 is a diagram showing a unit block of DCT processing in a frame processing mode / field processing mode in another specific format of a digital VTR. FIG.

FIG. 12 shows another example of FIG. 11. FIG.

13 illustrates a pair of macro blocks.

FIG. 14 is a diagram for explaining the state of processing in the frame processing mode in the example of FIG.

FIG. 15 is a diagram for explaining a state of processing in the field processing mode in the example of FIG.

FIG. 16 is a diagram for explaining a modification (with respect to previous prediction) of extension bit addition in the second embodiment. FIG.

17 is a block diagram showing the structure of a decoder.

18 shows an image of a radix cycle.

19 shows an image of even cycles.

20 is a diagram for explaining each predictive image.

21 is a diagram showing a data structure.

22 is a diagram showing a certain image of a moving object.

* Explanation of symbols for main parts of the drawings

10: frame memory root 11, 31: address generator

12 differential detector 13 DCT circuit

14 quantizer 15 variable length coding circuit

16: buffer 17: inverse quantizer

18 DCT circuit 20 Frame memory root with motion compensator

21: field motion detection circuit 22: frame motion detection circuit

23: motion prediction mode determination circuit 24: selector

25: Blocking mode determination circuit 43: Mode determination circuit

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a high efficiency coding apparatus for an image signal for high efficiency coding for an orthogonal transformation and a decoding apparatus thereof.

As a method of high efficiency encoding of a picture signal, for example, a standard of MPEG (Moving Picture Experts Group) defines a so-called high efficiency coding method of a picture signal for a digital storage medium. Here, the storage substitutes targeted by the scheme are those whose continuous transfer rate is about 1.5 Mbit / sec or less, such as so-called CD (compact disc), DAT (digital audio tape), hard disc, and the like. In addition, this is not only directly connected to a decoder, but also connected via a transmission medium such as a computer bus, a local area network (LAN), telecommunications, and the like. In addition, not only forward playback but also random access and high speed playback are also possible. Special features such as reverse playback, etc. are also considered.

The principle of the high efficiency coding method of the image signal by MPEG is as follows.

That is, in this high efficiency coding method, first, the redundancy in the time axis direction is lowered by taking the difference between the images, and then using the so-called discrete cosine transform (DCT) processing and variable length code, the space axis direction is used. To reduce the degree of redundancy.

First, the redundancy degree in the said time axis direction is demonstrated next. In general, in a continuous moving picture, it is very similar to the image before and after time and the image of attention (that is, image of a certain time). Therefore, for example, as shown in FIG. 20, if the difference between the image to be encoded from now and the image in front of it is taken and the difference can be transmitted, the redundancy in the time axis direction is reduced and transmitted. It is possible to reduce the amount of information. The picture to be encoded in this way is called a forward predictive coded picture (Predictive-Coded Picture, P picture or P frame) to be described later. Similarly, if the difference between the image to be encoded from now and the interpolated image made forward or backward or forward and backward in time is taken and the difference between the smaller values can be transmitted, It is possible to reduce the redundancy and reduce the amount of information to be transmitted. The picture to be encoded in this way is called a bidirectionally predicted picture (B picture or B frame) described later. In FIG. 20, the image indicated by I in the figure represents an intra-coded image (intra-coded picture, I image or I frame), which will be described later. An image is shown, and the image shown by B in the figure has shown the said B image.

In addition, so-called motion compensation is performed to produce each side image.

In other words, according to this motion compensation, for example, a block of 16 × 16 pixels (hereinafter referred to as a macro block) composed of unit blocks of 8 × 8 pixels is made, and the first position in the vicinity of the corresponding macro block position on the telephone is made. By searching for a case where the difference is small and taking the difference from the searched macro block, the data that must be sent can be reduced. In practice, for example, in the P picture (forward predictive coded picture), the difference between the prediction after the motion compensation and the difference between the prediction and the motion compensation after the motion compensation is small. 16 bits are selected and encoded in macroblock units.

However, in the case described above, for example, a large amount of data must be sent with respect to the portion (image) that comes after the object moves. Therefore, for example, in the B picture (bidirectional predictive coded picture), the interpolation picture created by adding the both of the temporally forward or backward picture after motion compensation already decoded and the picture to be encoded from now on. The difference and the one which does not take the difference, that is, the least amount of data among the four images to be encoded from now on are encoded.

Next, the redundancy degree of the said space axis direction is demonstrated next.

The copy of the image data is not transmitted as it is. Discrete cosine transform (DCT) is performed for each unit block of the 8x8 pixel. The DCT expresses the image in terms of how much frequency component of the cosine function is included instead of the pixel level. For example, data of a unit block of 8x8 pixels is stored in the two-dimensional DCT by the two-dimensional DCT. Is transformed into a coefficient block of an 8x8 cosine functional component. For example, a naturalized image signal captured by a TV camera often becomes a smooth signal. In this case, the data amount can be efficiently reduced by performing DCT processing on the image signal.

That is, for example, in the case of a smooth signal such as the above-described naturalized image signal, a large value is concentrated around a certain coefficient by performing the DCT. When the coefficient is quantized, the 8x8 coefficient block becomes almost zero, so that only a large coefficient can remain. Therefore, when data of this 8x8 coefficient block is transmitted, a so-called half-man code with a pair of so-called zero runs indicating how many nonzero zeros precede the coefficients in the order of so-called zigzag scans By making it possible to send, it is possible to reduce the amount of transmission. On the decoder side, the image is reconstructed in the reverse order.

Here, FIG. 21 shows a structure of data handled by the above-described encoding method. That is, the data structure shown in FIG. 21 is sequentially divided into block layers, macroblock layers, slice layers, picture layers, group of picture layers (G), and video sequence layers. It consists of. Hereinafter, in this FIG. 21, it demonstrates sequentially from the following layer.

First, in the block layer, the block of the block layer is composed of 8x8 pixels (pixels of 8 lines x 8 pixels) of which luminance or color difference is adjacent. The DCT (Discrete Cosine Transform) described above is performed for each unit block.

In the macro block layer, the macro block of the macro block layer is the same as the luminance block on the image of four luminance blocks (unit blocks of luminance) Y10, Y1, Y2, and Y3 adjacent to the left and right and up and down. It consists of six blocks with all of the color difference block (unit block of a color difference) Cr and Cb corresponding to a position. The order of such block transfer is Y0, Y1, Y2, Y3, Cr, Cb. Here, in the coding method, what is used for prediction (image of the reference which takes the difference), whether or not the difference should be sent is determined in this macroblock unit.

The slice layer is composed of one or a plurality of macro blocks arranged in the scanning order of images. In this slice head (header), the difference between the motion vector and the DC (direct current) component in the image is reset, and the first macro block has data indicating the position in the image, so that even if an error occurs, it is returned. I can do it. As a result, the slice length and the starting position are arbitrary and can be changed by an error state of a transmission path.

In the image layer, an image, that is, an image of one sheet is composed of at least one or a plurality of the slices. Each of the above-described intra coded pictures (I pictures or I frames), the forward-side coded pictures (P pictures or P frames), bidirectional predictive coded pictures (B pictures or B frames), and the like, according to the encoding method. It is classified into four types of pictures of DC intra coded picture (DC coded (D) picture).

Here, in the intra coded image (I image), only information ending in the middle of one image is used when the image is encoded. In other words, therefore, it is possible to reconstruct the picture only with the information of the I picture itself at the time of decoding. In practice, DCT processing is performed as it is without taking the difference. This coding method is generally inefficient, but if it is put everywhere, random access and high speed playback are possible.

In the forward predictive coded picture (P picture), an I picture or a P picture which has already been decoded in advance in time as an input is used as the predicted picture (the picture to be the difference). In practice, the macroblock units are selected to encode the difference from the motion-compensated predictive image, to encode the difference as it is without taking the difference (intra code), or to have higher efficiency.

In the bidirectional predictive coded picture (B picture), three types of pre-decoded I picture or P picture which are located temporally in advance and an interpolation picture made in both directions are used as the predictive picture. As a result, it is possible to select, in macroblock units, the most efficient one among the differential coding and intra code after the three types of motion compensation.

The DC intra coded image is an intra coded image composed only of the DC coefficient of the DCT, and cannot exist in the same sequence as the other three types of images.

The image group (GOP) layer is composed of one or a plurality of I images and zero or a plurality of non-I images. Here, the input order to the encoder is, for example, 1I, 2B, 3B, 4P * 5B, 6B, 7I, 8B, 9B, 10I, 11B, 12B, 13P, 14B, 15B, 16P * 17B, 18B, In the case of 19I, 20B, 21B, and 22P, the output of the encoder, i.e., the input of the decoder, is, for example, 1I, 4P, 2B, 3B * 7I, 5B, 6B, 10I, 8B, 9B, 13P, 11B, 12B, 16P, 14B, 15B * 19I, 17B, 18B, 22P, 20B, 21B. In this way, the order is changed in the code stream. For example, when the B picture is encoded or decoded, the I picture or P picture, which is backward in time, which becomes the prediction picture, is not encoded before. Because not. Here, the intervals (eg 9) of the I picture and the intervals (eg 3) of the I picture or the B picture are free. In addition, the space | interval of an I picture or a P picture may be changed inside the said image group layer. In addition, the interlayer of an image group layer is represented by said *. In addition, I is an I image, P is a P image, and B is a B image.

The video sequence layer is composed of one or a plurality of image group layers having the same image size, image rate, and the like.

As described above, in the case of transmitting a moving picture standardized by the high-efficiency coding method according to the MPEG, an image obtained by compressing one image in the image is sent first, and then the difference from the image which has motion-compensated this image. Is sent.

By the way, in the above one image, for example, when a field is processed as an image, since the vertical positions are alternately shifted in two fields, differential information must be transmitted even when transmitting still images, for example.

For example, when processing a frame as an image, an image shaken in a so-called comb shape must be processed for a moving part in the frame, for example. That is, for example, as shown in FIG. 22, when there is a fuselage CA such as a car in front of a stationary background, when one frame is observed, there is a movement between the fields. KS) burns.

Further, for example, in the case of processing an image in which a still part and a moving part are mixed, the compression efficiency in the image is reduced even if any method of processing the field as an image or processing a frame as an image is used. Bad parts are burned.

Therefore, the present invention has been proposed in view of the above-described circumstances, and it is possible to perform field processing or frame processing for a moving picture of a field structure even if the image has little motion or the image has many motions, or the image is a mixture of both. It is an object of the present invention to provide a high efficiency encoding device and a decoding device thereof which can be efficiently performed.

In order to solve such a problem, the 1st high efficiency coding apparatus which concerns on this invention is the high efficiency coding apparatus of the image signal which encodes the macroblock which consists of a two-dimensional array of several pixel as a unit, The said macroblock Means for detecting the difference between the motion vector between the frames and the absolute value of each pixel in units, and the difference between the absolute value of each pixel and the motion vector between the field, which is divided by the radix or even number of the pixel scan of the frame in the macroblock unit. Motion detection means comprising: a means for detecting motion; a frame prediction mode for performing motion compensation in units of frames in the macroblock; and a field prediction mode for performing motion compensation in units of fields in the macroblocks. Recall that it is effective in compensating for motion A first mode selecting means for judging by the information output from the motion detecting means, for selecting an efficient prediction mode, and a frame processing mode and the macro for blocking the frame in the macroblock as an orthogonal transform as a unit; It is determined by using the information output from the motion detecting means and the first mode selecting means which of the field processing modes that block the field in the block, such as performing orthogonal transformation as a unit, is effective in performing orthogonal transformation. Second mode selecting means for selecting an efficient blocking mode, and an odd cycle of a period in which a radix field is scanned in an interlace of an encoding process for one frame or an even cycle in a period in which an even field is scanned? In the corresponding radix cycle Address generating means for controlling a frame memory group so as to output a blocked macroblock corresponding to the mode of blocking, motion prediction mode information selected by the first mode selecting means, and selected by the second mode selecting means It is equipped with a motion compensation means for receiving the blocking mode information, and performing the motion compensation frame or inter-field prediction corresponding to the mode information.

A second high efficiency encoding apparatus according to the present invention is a high efficiency encoding apparatus of an image signal which encodes a macro block composed of a two-dimensional array of a plurality of pixels as a unit. Means for detecting the difference between the absolute value of the vector and each pixel, and means for detecting the difference between the absolute value of each pixel and the motion vector between the fields, each of which is divided by the radix or even number of the scan of the pixel of the frame in units of the macroblock. Motion compensation means comprising a motion detection means comprising: a frame prediction mode for performing motion compensation in units of frames in the macroblock, and a field prediction mode for performing motion compensation in units of fields in the macroblock. Output from the motion detecting means The first mode selection means for selecting an efficient prediction mode, the frame processing mode for blocking the frame in the macroblock as a unit, or the like, and the field in the macroblock. Which of the field processing modes that block to perform orthogonal transformation as a unit is effective in performing orthogonal transformation using the information output from the motion detecting means and the first mode selecting means to determine the effective mode of blocking. Recognizing whether the second mode selection means for selecting and the even cycle of the period of scanning the odd field or the period of scanning the excellent field in the interlace of the encoding process for one frame, Only when the mode is field processing mode, every As outputs one frame in turn, the odd field block of the time. Subsequently, address generation means for controlling a frame memory group so as to output one even field of a macroblock in order in the even cycle, the motion prediction mode information selected by the first mode selection means, and the second mode. And a motion compensation means for receiving the block mode information selected as the selection means and performing the motion compensation frame or the inter-field prediction corresponding to the mode information.

In addition, the high-efficiency decoding apparatus according to the present invention receives and decodes the encoded picture data to be reproduced, and header information including the detected motion vector information, the motion prediction mode information, and the blocking mode information, and together with the decoded picture decoded data. Inverse-variable encoding means for outputting macroblock address increments in said detected motion vector information, motion prediction mode information, blocking mode information, and said header information of a macroblock, and said image decoded data from said macroblock address increments; Calculates an address increment value in a frame buffer which accumulates a value, obtains a head address of the macroblock, and gives address heading means for giving the head address to the frame buffer, and the relative addresses of the macroblocks other than the head address. remind In addition to the frame buffer, the data is accessed, the detected motion vector, the motion prediction mode information, and the blocking mode information are received, and motion compensation frames or fields corresponding to the mode information are subjected to prediction to perform motion compensation. Motion compensation means for constructing the frame buffer.

According to the present invention, the frame processing mode and the field processing mode can be switched in units of macro blocks, and therefore, the most efficient encoding can be selected in units of macro blocks.

EXAMPLE

EMBODIMENT OF THE INVENTION Hereinafter, the Example which applied this invention is described, referring drawings.

1 and 2 show an embodiment of a high efficiency coding apparatus of an image signal according to the present invention, and FIG. 17 is a system diagram showing an embodiment of a high efficiency decoding apparatus of an image signal according to the present invention.

FIG. 1 shows a first embodiment of the present invention, in which a macroblock consisting of a two-dimensional array of pixels smaller than one screen (e.g., 16 x 16 pixels in a spatial arrangement of input image data in raster scan order) A high efficiency encoding device for encoding an image signal in units of blocks, wherein a plurality of frames (one screen) consisting of a plurality of unit blocks of 16 × 16 pixels are stored in a plurality of frames and stored as an original picture Frame motion detection, which is a motion detection means for detecting the motion vector between the memory root 10 and the field divided by the radix or even number of the scan of the frame pixel between the frame and the macroblock unit, and the absolute value difference of each pixel. The circuit 22 and the field motion detection circuit 21 perform motion compensation in units of frames in the macro block. Determines whether the efficiency of the frame prediction mode or the field prediction mode that performs motion compensation on the basis of the field in the macroblock is good, based on the information output from the motion detection means, and selects an efficient prediction mode. The motion prediction mode determination circuit 23 and the selector 24, which are the first mode selection means, and a frame processing mode for blocking orthogonal transformation as a frame unit in the macroblock, and a field in the macroblock as a unit. As a result, it is determined by using the information output from the motion detecting means and the first mode selecting means, which of the field processing modes to block orthogonal transformation is effective in performing orthogonal transformation, so that an efficient blocking mode is obtained. Blocking mode determination circuit 25, which is a second mode selection means for selecting a. And whether or not the odd cycle of the period in which the radix field scan is performed or the even cycle in the period in which the even field scan is performed in the interlace of the encoding process for one frame (one screen) is performed. An address generator 11 which is an address generating means for controlling a frame memory root to output a correspondingly blocked macroblock, and a motion prediction mode information selected by the first mode selecting means and a blocking selected by the second mode selecting means A frame memory group 20 with a motion compensator, which is a motion compensation means for receiving mode information and performing motion compensation frames or inter-field prediction in response to the mode information.

2 shows a second embodiment of the present invention. Moreover, in Fig. 2, blocks designated by the same reference numerals as in Fig. 1 perform the same function. Therefore, a block numbered different from that in FIG. 1 will be described here. That is, in the high efficiency encoding apparatus of FIG. 2, in addition to the blocks numbered the same as the high efficiency encoding apparatus of FIG. 1, the motion compensation is performed when the orthogonal transform is blocked in the frame prediction mode and the motion compensation is performed. Mode which is a mode selection means which selects the efficient prediction mode by judging by the information output from the said motion detection means which efficiency of the block of orthogonal transform in the said field prediction mode is the case of the said field processing mode. Whether the decision circuit 43, the selector 24, and one frame (one screen) is an odd cycle of a period in which the radix field is scanned in the interlace of the encoding process or an even cycle in the period in which the even field is scanned Recognizes that macros in the odd cycle are only available when the mode of the mode determining circuit 43 is field prediction field processing. The address generator 31 is provided as an address generating means for controlling the frame memory group so that the odd number field of the lock is sequentially outputted one frame, and then the even field of the macroblock is outputted one frame in order in the even cycle.

Further, the second embodiment is an encoding device without the partitioning mode and the mode division of the motion compensation. Of course, the same block diagram as that of the first embodiment may be used, but the second embodiment is fundamentally different from the first embodiment in the operation of the address generator as described above.

First, the main flow of the image data subjected to the encoding process in the first embodiment will be described using the configuration of FIG. Next, a second embodiment will be described with reference to FIG.

That is, in FIG. 1, a digital image signal is supplied to the input terminal 1, and is stored in the frame memory group 10. As shown in FIG. From the frame memory group 10, data of the unit macroblock of the 16x16 pixel is controlled and output to the address generator 11 described later, and transmitted to the difference detector 12. The difference detector 12 is also supplied with motion compensated image data from the frame memory group 20 with a motion compensator to be described later, and these differences are detected by the difference detector 12.

The output of the difference detector 12 is sent to a DCT circuit 13 that performs orthogonal transform (DCT) processing. The DCT coefficient data obtained by DCT processing in the DCT circuit 13 is sent to the quantizer 14. The quantized data from the quantizer 14 is output via the variable length coding circuit 15 and the buffer 16, which perform variable length coding processing such as so-called half-man coding and Langlenth coding. It is output as encoded data from (2).

In the frame memory group 20 with the motion compensator, the quantization data from the quantizer 14 includes an inverse quantizer 17 that performs inverse quantization processing of the quantization processing in the quantizer 14; The data via the adder 19 can be supplied via the reverse DCT circuit 18 which performs the reverse DCT process of the DCT process in the DCT circuit 13. In the adder 19, the output of the inverse DCT circuit 18 and the output of the frame memory group 20 with the motion compensator are added. In addition, from the buffer 16, a signal for preventing the overflow of the buffer 16 can be fed back to the quantizer 14.

On the other hand, image data output from the frame memory group 10 in units of the macro blocks are transmitted to the frame motion detection circuit 22 and the field motion detection circuit 21.

The frame motion detection circuit 22 detects the absolute difference between the motion vectors between the frames and the pixels in units of the macroblock, and determines their data (the data FMMV of the motion vectors between the frames and the absolute value difference data FMAD). The field motion detection circuit 21 detects the absolute difference between the motion vectors between the fields and each pixel in units of macroblocks, and compares the data (FDMV) and the absolute value difference between the motion vectors between these data fields. The data FDAD is outputted The data FMMV / FDMV of each motion vector of these motion detection circuits 21 and 22 is transmitted to the selector 24, and the data FMAD / FDAD of each absolute value difference summation is described above. It is sent to the motion prediction mode determination circuit 23.

The motion prediction mode determination circuit 23 is based on the absolute value difference data FAD from the frame motion detection circuit 22 and the absolute value difference data FDAD from the field motion detection circuit 21. In the motion prediction processing in the frame memory group with a motion compensator 20 to be described later, it is determined whether the motion prediction processing is performed in the frame unit or the motion prediction processing in the field unit. Output data indicating the processing mode of the more efficient one). Specifically, in this motion prediction mode determination circuit 23, for example, the difference between the absolute value difference data FMAD and the absolute value difference data FDAD is greater than any threshold value T1 (FMAD-FDAD> T1). Is determined, outputting data (field data MPED in the field processing mode in the motion prediction) indicating that the motion prediction processing is performed from the circuit 23 on a per field basis. When it is determined that the difference between the absolute difference data FMAD and the absolute difference data FDAD is less than or equal to the threshold value T1 (when FMAD-FDAD≤T1), the motion prediction is performed on the frame basis. Outputs data indicating that the processing is more efficient (data of the frame processing mode (MPFM) in motion prediction.) Either of these motion-side mode data (MPFM / MPFD). Is sent to the frame memory group 20 with a motion compensator, whereby the frame memory group 20 performs motion compensation on a frame basis or field basis, and these motion prediction mode data (MPFM / MPFD). Is also sent to the selector 24.

The selector 24, according to the motion prediction mode data (MPFM / MPFD) from the motion prediction mode determination circuit 23, motion vector data (FMMV) between frames supplied from the frame motion detection circuit 22. And either of the motion vector data FDMV between the fields supplied from the field motion detection circuit 21 is output. That is, when the motion prediction mode data is the data MPFD indicating the field prediction mode, the motion vector data FDMV from the field motion detection circuit 21 is selected and output, and the motion prediction mode data is frame prediction. In the case of the data MPFM indicating the mode, the motion vector data FMMV is selected from the field motion detection circuit 22 and output. The motion vector data FMMV / FDMV selected by the selector 24 is sent to the blocking mode determination circuit 25.

The block mode determination circuit 25 is also supplied with output data from the frame memory group 10 and the processing mode data MPFM / MPFD from the motion prediction mode determination circuit 23. The block mode determination circuit 25 receives the motion prediction mode data (MPFM / MPFD) and the motion vector data (FMMV / FDMV) and makes a difference using an image from the frame memory group 10. An image is created and the block processing mode most suitable for the image output from the frame memory group 10 and subjected to DCT processing by the DCT circuit 13 is selected based on the difference image. Further, in the case of the above I image (or I frame), the image (original) data of the frame memory group 10 is used instead of the difference image.

In other words, suppose that the macroblock in the difference image is, for example, a macroblock as shown in FIG. 3 (macroblock of the original picture in the I picture). In Fig. 3, the odd lines o1, o2, o3, ..., oN, where N is 16 in the case of a macro block, are indicated by solid lines, and even lines (e1, e2, e3, ..., eN, In the case of macroblock, N is indicated by dotted line. Each pixel of the even line is represented by e (i, j), and the odd line is represented by each pixel as o (i, j). In the difference image or the original image (the image of the I image) as shown in FIG. 3, the difference EFD of the difference image in the field unit can be expressed by a number 1 expression, and the difference (EFM) of the difference image in the frame unit. ) Can be represented by the formula of number 2.

[1]

[Number 2]

In the blocking mode determination circuit 25, the difference between the difference EFM obtained in the frame and the difference EFD obtained in the field using the formulas of numbers 1 and 2 is more than any threshold value T2. In the case where it is determined to be large (when EFM-EFD> T2), data (data of the field processing mode (MPED) in the block processing) indicating that the DCT in the DCT circuit 13 is performed in units of fields is output. Conversely, when it is determined that the difference between the difference EFM and the difference EFD is less than or equal to the threshold T2 (when EFM-EFD? T2), the DCT in the DCT circuit 13 is determined. Is output in the frame unit (data MDFM in the frame processing mode in the block processing). Any of these block mode data MDFM / MDFD is transmitted to the address generator 11 and the frame memory group 20 with a motion compensator. The motion vector data FMMV / FDMV, the block mode data MDFM / MDFD, and the prediction mode data MPFM / MPFD are sent to the variable length coding circuit 15 described above.

In the address generator 11, for example, the image data stored in the frame memory group 10 is blocked in accordance with processing mode data (MPFM / MPFD) in the DCT in units of the macro blocks. The frame memory group is controlled to output a block. In other words, in the address generator 11, in the case of the data MPFM in which the blocking mode data indicates DCT processing on a frame basis, as shown in FIG. 4, the macroblocks in which even and odd numbers are alternately scanned are output. The frame memory group is controlled. As a result, the unit block of the macro block sent to the DCT circuit 13 is the sum of the even field and the odd field. Conversely, in the case of the data MDFD in which the blocking mode data indicates DCT processing on a field-by-field basis, as shown in FIG. 5, the frame memory is configured to output the scanned macroblock by dividing the even and odd scans separately. Control your army. As a result, the unit block of the macro block sent to the DCT circuit 13 is obtained by dividing the even field and the odd field separately. In the DCT circuit 13, however, DCT conversion is performed in a unit block of 8x8 pixels as described above. In Figs. 4 and 5, the odd lines are shown by solid lines, and the even lines are shown by dotted lines.

Further, in the frame memory group 20 with the motion compensator, prediction mode data (MPFM / MPFD) from the motion prediction mode determination circuit 23 and blocking mode data (MPFM) from the blocking mode determination circuit 25 are included. / MPFD and the motion vector data (FMMV / FDMV) selected by the selector 24 are supplied.

Therefore, in the frame memory group 20 with the motion compensator, the motion vector corresponds to the prediction mode data MPFM / MPFD in the motion prediction and the blocking mode data MDFM / MDFD in the DCT process. Motion compensation using data FMMV / FDMV is performed.

By the way, the mode determination circuit 43 in FIG. 2 has absolute value difference data FMAD from the frame motion detection circuit 22 and absolute value difference data from the field motion detection circuit 21. According to FDAD), the motion prediction process is performed in the frame unit or the field prediction process in the frame prediction group in the frame memory group 20 with a motion compensator to be described later. The determination result (corresponds to the prediction mode data (MPFM / MPFD) in the first embodiment), the motion vector (FMMV.FDMV) from the motion detection circuits 21 and 22, and the frame memory group ( A difference image is created using the image from 10), and is output from the frame memory group 10 based on the difference image, and is most suitable for the DCT processed image in the DCT circuit 13. And it determines a mode of rokhwa treatment. That is, in the mode determination circuit 43, which of the block prediction mode processing mode PDFM in motion prediction is the frame prediction mode and the block prediction field processing mode PDFD in motion prediction is the most efficient? Is being judged. In other words, the mode determining circuit 43 has the same configuration as the combined functions of the motion prediction mode determining circuit 23 and the blocking mode determining circuit 25 in the first embodiment.

Further, the determination of the specific mode can be performed, for example, in the same manner as the determination of the motion prediction mode and the blocking mode in the first embodiment.

In addition, the address generator 31 outputs a macroblock blocked according to the mode data (PDFM / PDFD) in the macroblock units, for example, on the image data stored in the frame memory group 10. The frame memory group 10 is controlled to do so. That is, in the address generator 31, in the case of the data (PDFM) indicating the coding processing in units of frames, the mode generator outputs a macroblock in which even and odd numbers are alternately scanned as shown in FIG. The frame memory group 10 is controlled. As a result, the unit block of the macro block sent to the DCT circuit 13 is the sum of the even field and the odd field. Conversely, in the case of the data (PDFD) indicating the encoding processing in field units, the mode data outputs the odd field for the macroblock for one frame (one screen) in the odd cycle, followed by the even cycle. The frame memory group 10 is controlled to output the even field for the macroblock for one frame (one screen) in sequence. As a result, the unit block of the macro block sent to the DCT circuit 13 in the odd cycle becomes a macro block composed of only the odd field, and becomes a macro block composed of only the even field in the even cycle. In the DCT circuit 13, however, DCT conversion is performed in a unit block of 8x8 pixels as described above. (In addition, in Figs. 4 and 5, the odd lines are represented by solid lines, and the even lines are Indicated by a dashed line).

That is, in the high-efficiency encoding apparatus of the image signals of the first and second embodiments described above, the frame prediction mode and the field prediction mode in motion prediction, and the frame processing mode and the field processing mode in the block of DCT processing are selected. Since it can be changed in units of macroblocks, the most efficient encoding is possible in the macroblock units.

Specifically, the encoding apparatuses of the first and second embodiments perform motion prediction and DCT conversion processing as shown below for each format of a so-called digital VTR, for example.

Here, in FIGS. 6, 8, and 10, the field Io field (the radix field of the I frame) and the Ie field (the even field of the I frame) constituting the frame of the I frame (I image) are assumed. A field constituting the P frame (P picture) is a Po field (odd field) and a Pe field (excellent field), and a field constituting the B frame (B picture) is a Bo field (9 odd field) and a Be field. (Excellent field).

In addition, in the first and second embodiments, as shown in FIG. 4, the frame processing mode in block formation constitutes the macroblock by adding the odd field and the even field (that is, forming a macroblock for each frame). This macroblock is a processing unit, and as shown in FIG. 5, the field processing mode in block formation constitutes a macroblock separately in the radix field and the even field (that is, for each field). Macroblock), which is a macroblock processing unit. Thus, for example, in an I frame, a frame processing mode and a field processing mode are switched for each macroblock.

In the high-efficiency encoding apparatus of the first and second embodiments, the odd cycle of the period in which the radix field in the interlace is scanned by the encoding process and the even period in which the even field is scanned for one frame It is divided into cycles.

In the case of the first embodiment, for example, when dealing with a so-called digital VTR format of 4: 2: 0 components, as shown in FIG. 7, when the blocking is in the frame processing mode, the luminance consists of the odd field and the even field. DCT processing is performed on each unit block of the macro block consisting of the blocks Y0, Y1, Y2, Y3 and the color difference blocks Cb0, Cr1 of the odd field. On the other hand, in the blocker field processing mode, the block comprises the luminance blocks Y02o and Y13o of each radix field, the luminance blocks Y02e and Y13e of each even field and the color difference blocks cb0 and cr1 of the hold field. DCT processing of each unit block of the macro block MB is performed.

As shown in FIG. 8, the motion prediction in the example of FIG. 7 enables motion prediction (MCP) between an I frame and a P frame in the frame prediction mode. On the other hand, in the field prediction mode, the motion prediction between the Io field and the Po field (MCoPo), the motion prediction between the Io field and the Pe field (MCePo), and the motion prediction between the Ie field and the Pe field (MCe Pe) It becomes possible. That is, in FIG. 8, motion prediction and blocking may exist independently in the frame prediction / processing mode and the field prediction / processing mode, and one motion vector is obtained in the frame prediction mode, and the motion vector is obtained in the field prediction mode. Two are saved.

Therefore, in the first embodiment, for example, when the blocking of an I frame is a frame processing mode, in the radix cycle, the Io field and the Ie field are combined to form the macro block, for example the In the odd cycle, DCT conversion (where DCT is performed for each of the unit blocks of 8x8), quantization, and variable subcomplementation are performed for each macroblock. On the other hand, none of the even cycles in this mode send data.

In addition, when the blocking is in the field processing mode, the macroblock is configured in a form in which the Io field and the Ie field are divided separately in the radix cycle, and DCT conversion (for DCT is 8 × 8) for each macroblock. Per unit block), quantization, and variable length coding are performed. In contrast, in the even cycle of this mode, as can be seen from FIG. 7, none of the data is sent.

In the case of the P frame, the following processing is performed. For example, when the blocking of the P frame is the frame prediction mode and the motion prediction is the frame prediction mode, the radix cycle detects the reference picture as the omnidirectional picture (picture 0 of the I frame) and the motion vector (MVP) between frames. Then, the difference between the macroblock in which the Io field and the Ie field are alternately combined is encoded as a predictive image, and the difference between the original image and the original image is not transmitted.

In addition, when the block of the P frame is the motion prediction mode in the frame processing mode, in the odd cycle, the Io field and the Ie field (or the Po field and the Pe field) are referred to as reference pictures, respectively, between the Io field and the Po field. Motion vector (MVoPo), motion vector (MVePo) between Ie and Po fields, motion vector (MVoPe) between Io and Pe fields, motion vector (MVe Pe) between Ie and Pe fields, and detect Among the field prediction, the even field prediction, and the bidirectional prediction (eg, the average of the good field prediction and the odd field prediction), the prediction with the minimum prediction error with the current P frame is selected, so that the Io field and the Ie field are selected. The combined macroblock is encoded as a difference from the original picture as a predictive picture. On the other hand, none of the even cycles in this mode send data.

Further, when the blocking of the P frame is in the field processing mode and the motion prediction is in the frame prediction mode, the motion cycle (MVP) between the frames is detected as the reference picture as the picture of the I frame (or the picture of the P frame) in the odd cycle. Then, the difference between the original block (the macroblock formed by dividing the Po field and the Pe field separately) is encoded as the predicted image using the macroblock formed by dividing the Io field and the Ie field separately. On the other hand, none of the even cycles in this mode send data as described above.

Further, when the blocking of the P frame is in the field processing mode and the motion prediction is in the field prediction mode, the Io field and the Ie field (or the Po field and the Pe field) are referred to as reference pictures in the odd cycle, respectively, between the Io field and the Po field. Detects the motion vector MVoPo, the motion vector MVePo between the Io field and the Po field, the motion vector MVoPe between the Ie field and the Pe field, and the motion vector MVe Pe between the Ie field and the Pe field, From the prediction of the odd field, the prediction of the even field, and the prediction of both directions (for example, the average of the prediction of the even field and the odd field prediction), the prediction with the minimum prediction error with the current P frame is selected. The difference between the original block (the macroblock formed by dividing the Po field and the Pe field separately) is encoded as the predicted image using the macroblock formed by dividing the fields separately. On the other hand, none of the even cycles in this mode send data.

In the case of the B frame, the following processing is performed.

For example, when the blocking of the B frame is the frame prediction mode and the motion prediction is the frame prediction mode, in the radix cycle, the reference picture is the front and back picture as the motion vector between the frames, that is, between the I frame and the B frame. Detects the motion vector (FMUB) and the motion vector (BMVB) between the P and B frames, and among the forward prediction, backward prediction and bidirectional prediction (average of forward prediction and backward prediction), the prediction error with the current frame is minimal. The prediction is selected, and the difference between the original image and the original block is encoded using the macroblock in which the odd field and the even field are alternately combined. On the other hand, none of the even cycles in this mode send data.

Further, when the motion of the B frame is the frame prediction mode and the motion prediction is the field prediction mode, in the radix cycle, prediction of the odd field and prediction of the even field are performed for these pictures as the reference picture as the front and rear pictures, respectively. Each motion vector, i.e., the motion vector between the Io field and the Bo field (FMVoBo), the motion vector between the Ie field and the Bo field (FMVeBo), the motion vector between the Io field and the Be field (FMVoBe), and Ie. Motion vector between field and Be field (FMVeBe), motion vector between Po field and Bo field (BMVoBo), motion vector between Pe field and Bo field (BMVe Bo), motion vector between Po field and Be field (BMVoBe ), The motion vector (BMVeBe) between the Pe field and the Be field is detected, and the prediction of the odd field and the prediction of the even field and both predictions (e.g., the prediction of the even field and the odd field prediction) Rating ) From, by selecting the prediction is the prediction error of the current frame is minimized, and the Io field and the Ie field (or the Po field and the Pe field) are combined the macroblock as a predicted image coding a difference from the original image. On the other hand, none of the even cycles in this mode send data.

Further, when the blocking of the B frame is the field prediction mode and the motion prediction is the frame prediction mode, in the cardinal cycle, the reference picture is the front and rear picture, and the motion vector between the frames, that is, the motion vector between the I frame and the B frame. (FMVB) and a motion vector (FMVB) between a P frame and a B frame, and among prediction between the forward prediction and the backward prediction bidirectional prediction (average of the forward prediction and the backward prediction), the prediction error with the current frame is minimized. Is selected to encode the difference with the original picture as a predictive picture of the macroblock formed by dividing the odd field and the even field separately. On the other hand, none of the even cycles in this mode send data.

When the motion of the B frame is the field prediction mode and the motion prediction is the field prediction mode, the odd cycle predicts the odd field and the even field, respectively, for these pictures as reference pictures as the front and rear pictures. Is done. Each motion vector, that is, a motion vector between Io and Bo fields (FMVoBo), a motion vector between Ie and Bo fields (FMVeBo), a motion vector between Io and Be fields (FMVoBe), Ie and Be fields Motion vector between fields (FMVeBe), motion vector between Po field and Bo field (BMVoBo), motion vector between Pe field and Bo field (BMVeBo), motion vector between Po field and Be field (BMVoBe), Pe field Detects the motion vector (BMVeBe) between and Be fields and predicts the radix field by each vector, the prediction of the even field, and both predictions (e.g., the average of the prediction of the even field and the prediction of the radix field). Selects a prediction having a minimum prediction error with the current frame, and encodes the difference from the original picture as a predictive picture of the macroblock formed by dividing the Io field and the Ie field (or the Po field and the Pe field) separately. . On the other hand, none of the even cycles in this mode send data.

However, in the first embodiment, as can be seen from FIG. 8, the motion prediction between the Io field and the Ie field, the motion prediction between the Po field and the Pe field, and the motion prediction between the Bo field and the Be field are not possible. .

In this case, using the second embodiment, prediction can be made from the odd field to the even field in each image. That is, for example, as shown in FIG. 9, when the blocking is in the frame processing mode, the color difference between the luminance blocks Y0, Y1, Y2, and Y3 consisting of the odd field and the even field in the odd cycle and the odd field DCT processing of each unit block of the macro block MB made up of the blocks Cb0 and Cr1, and when the blocking is in the field processing mode, each luminance block Y02o, Y13o of the radix field in the radix cycle. ) And each unit block between the color difference blocks Cb0 and Cr1 of the radix field are subjected to DCT processing. Thereafter, each unit block of each of the luminance blocks Y02e and Y13e of the even field in the even cycle is subjected to DCT processing.

In the example of FIG. 9, the motion prediction includes the motion prediction between the Io field and the Io field (SMCI) in addition to the motion predictions (MVP, MCoPo, MCope, MCePo, MCePe) of FIG. The motion prediction between the Po field and the Pe field is possible.

Thus, in the second embodiment, for example, when the blocking of an I frame is a frame processing mode, in the odd cycle, the Io field and the Ie field are combined to form a fraudulent macro block. In an odd cycle, DCT conversion is performed for each of the macroblocks (however, DCT is performed for each unit block of 8x8). Quantization and variable length coding are performed. In contrast, none of the even cycles in this mode send data. When the blocking is in the field processing mode, only the odd field of the macroblock is encoded in the odd cycle. As a result, for example, at the end of the odd cycle, the decoder side, which will be described later, obtains the macroblock portion of the front of the Io field and the Ie field by the frame processing mode. Further, in the even cycle of the I frame, motion prediction is performed on the macro block of the Ie field in the field processing mode as a reference picture, and the difference picture between the motion vector SMVI and the predictive picture is encoded. .

In the case of the P frame, the following processing is performed. For example, when the blocking of the P frame is the frame prediction mode and the motion prediction is the frame prediction mode, in the odd cycle, the reference picture is the omnidirectional image (the picture of the I frame), and the motion vector (MVP) between the frames is selected. The difference and the difference from the original image are encoded as the predicted image using the macroblock in which the Io field and the Ie field are combined. On the other hand, no data is sent in the even cycle of this mode as described above.

Further, when the block is in the field processing mode and the motion prediction is in the field prediction mode, in the radix cycle, the Io field and the Ie field (or the Po field and the Pe field) are referred to as reference pictures, respectively, and the motion between the Io field and the Po field is obtained. Vector (MVoPo) and motion vector (MVe Po) between the Io field and the Po field are detected and, among the odd field prediction, even field prediction, and both predictions (e.g., average of the even field prediction and the odd field prediction), The prediction in which the prediction error with the odd field of the current frame is minimum is selected, and the difference with the prediction image is encoded. On the other hand, in the even cycle of this mode, for the macroblock of the field processing mode, the motion vector MVoPe between the Io field and the Pe field and the motion vector MVePe between the Ie field and the Pe field, and the Po field and Detects the motion vector (SMVP) between the Pe fields, predicts the odd field by each vector, predicts the even field, predicts the odd field of the current frame (predicts the motion from the Po field which performs only the good cycle), and among them. Prediction in which the prediction error is minimized is selected between the prediction by the two prediction averages, and the difference with the prediction image is encoded.

Also, for example, when the blocking of the B frame is the frame prediction mode and the motion prediction is the frame prediction mode, in the radix cycle, the reference picture is the front and rear picture as the motion vector between the frames, i.e. between the I frame and the B frame. The motion vector (FMVB) and the motion vector (BMVB) between P and B frames are detected, and among the forward prediction, backward prediction and bidirectional prediction (average of forward prediction and backward prediction), the prediction error with the current frame is minimal. The prediction to be selected is selected and the difference with the prediction image is encoded. On the other hand, no data is sent in the even cycle of this mode.

Further, when the block is in the field processing mode and the motion prediction is the field prediction mode, the reference pictures are forward and backward in the odd cycle, the odd field prediction and the even field prediction are performed on these pictures, respectively, and the respective motion vectors. That is, the motion vector between the Io field and the Bo field (FMVoBo), the motion vector between the Ie field and the Bo field (FMVeBo), the motion vector between the Po field and the Bo field (BMVoBo), and the motion vector between the Pe field and the Bo field ( BMVeBo) is detected. In the same manner as described above, the prediction with the minimum prediction error is selected, and the difference from the prediction image is encoded. In addition, in the excellent cycle of this mode, the motion vector (FMVoBe) between the Io field and the Be field, the motion vector (FMVerBe) between the Ie field and the Be field, the motion vector (BMVoBe) between the Po field and the Be field, and the Pe field Each prediction by the motion vector (BMVeBe) between the and Be fields, and also the prediction of the radix field of the current frame (that is, the prediction by the motion vector (SMVB) between the Bo field Be fields) is performed. A prediction is selected and the difference with the prediction image is encoded.

Furthermore, in the first embodiment, for example, when dealing with the so-called digital VTR format of 4: 2: 2 components, as shown in FIG. 11, the radix field and even field in the frame processing mode. DCT processing is performed on each unit block of the macroblock consisting of luminance blocks Y0, Y1, Y2, Y3 consisting of the color blocks and color difference blocks Cb01, Cr01, Cb23, Cr23 composed of odd and even fields. In the field processing mode, the luminance blocks Y02o and Y13o of the radix field and the color difference blocks Cb012o and Cr0123o of the radix field, the luminance blocks Y02e and Y13e of the even field and the color difference blocks of each even field ( DCT processing of each unit block of the macroblock consisting of Cb0123e and Cr0123e) is performed.

The motion prediction in the case of the example of FIG. 11 is as shown in FIG. 8 described above. However, in the case of FIG. 11, as described above, the motion prediction between the Io field and the Ie field, the motion prediction between the Po field and the Pe field, and the motion prediction between the Bo field and the Be field are performed. Can not.

In this case, therefore, as described above, the second embodiment may be used. That is, for example, as shown in FIG. 12, when the blocking is in the frame processing mode, the luminance blocks Y0, Y1, Y2, Y3 and the chrominance blocks Cb01, Cr01, Each unit block of the macroblock consisting of Cb23 and Cr23 performs DCT processing, and when the blocking is in the field processing mode, each of the luminance blocks Y02o and Y13o of the radix field and each of the radix field in the radix cycle. Each unit block with the color difference blocks Cb0123o and Cr0123o is subjected to DCT processing. Thereafter, DCT processing is performed on each unit block of each of the luminance blocks Y02e and Y13e of the even field and each of the chrominance blocks Cb0123e and Cr0123e of the even field in the even cycle.

In the case of this example of FIG. 12, the motion prediction is similar to that of FIG.

Further, in the first and second embodiments, when dealing with the digital VTR format of the 4: 2: 2 component, for example, in addition to the processing shown in FIGS. 11 and 12 described above, for example, in FIG. As shown, the frame motion prediction is performed in units of macroblocks (MB), but in the case of field motion prediction, a macroblock MB (i, j) and a macroblock MB (i) positioned below it are used. With + 1, j) as a pair, it is possible to perform the motion prediction of the odd field and the motion prediction of the even field for the pair MBg of the macroblock.

In the case of this FIG. 13 example, FIG. 14 shows that a macroblock of a part of the frame is extracted. The processing proceeds in the direction of the arrow in the figure of FIG. That is, in Fig. 14, the next macro block MB (i, j + 1) for a certain macro block MB (i, j) and the macro block MB (located on the next line) below them are shown. i + 1, j) and MB (i + 1, j + 1).

In the macroblock as shown in FIG. 14, for example, in the frame processing mode, each macroblock MB (i, j), MB (i, j + 1) ... MB (i + 1) For each MB (i + 1, j + 1), each luminance block (Y0, Y1) ALC chrominance blocks (Cb01, Cr01) are DCT-processed. This processing does not affect the processing mode of other macro blocks.

On the other hand, in the case of the field processing mode, as shown in FIG. 15, for the pair MBg of the macroblocks, the macroblocks constituting the pair MBg of the macroblocks are divided into the macroblocks of the odd field ( MBgo) and macroblocks MBge of even fields, and each luminance block Y0o, Y1o and chrominance blocks Cb01o, Cr01o in macroblock MBgo of the odd field are subjected to DCT processing. Here, for example, if the pair MBg of the macroblock is composed of the macroblocks MB (i, j) and MB (i + 1, j) shown in FIG. 14, the macroblock MBg The luminance blocks Y0o and Y1o in the macroblock MBgo of the radix field in the subfields are the luminance blocks of the radix field of the macroblock MB (i, j) and the macroblock MB (i + 1, j). The color difference blocks Cb01o and Cr01o in the macroblock MBgo of the radix field are the same as the color difference block of the radix field of the macroblock MB (i, j). It consists of a color difference block of the radix field of the macro block MB (i + 1, j). From the same, the luminance blocks Y0e and Y1e in the macroblock MBge of the even field are the luminance blocks of the even field of the macroblock MB (i, j) and the macroblock MB (i + 1). j)), and the color difference blocks Cb01e and Cr01e in the macroblock MBge of the even field are the color difference blocks of the even field of the macroblock MB (i, j). And a color difference block of even fields of the macro block MB (i + 1, j).

From the above, the relationship between the motion prediction and each processing mode of the DCT transform is described below. That is, in the encoding apparatus of the present embodiment, for example, when the DCT transformation of the frame processing mode is performed in the motion prediction of the frame processing mode for the macro block MB (i, j), for example, The image encoded in the frame memory group 20 with a motion compensator is used as a reference frame, and the difference between the predicted image extracted from the reference frame and the input image (original image) is DCT transformed. The DCT coefficient and the frame motion vector are transmitted.

Further, for example, in the macro block MB (i, j), in the motion prediction of the field processing mode, in the case of DCT transformation of the field processing mode, in the macro block MB (i, j), The difference between the predicted image extracted from the radix field and the original image of the radix field, and the motion vector of the radix field are encoded. In the macro block MB (i, j), the difference between the prediction image extracted from the even field and the original image of the even field and the motion vector of the even field are encoded.

For example, in the macro block MB (i, j), in the motion prediction of the field processing mode, in the DCT transformation of the frame processing mode, in the macro block MB (i, j), The frame difference between the predicted picture and the input picture of the position of the macroblock MB (i, j) extracted from the reference frame, the motion vector of the odd field, and the motion vector of the even field are transmitted. Further, in the macro block MB (i + 1, j), the frame difference between the predicted picture and the input picture of the position of the macroblock MB (i + 1, j) extracted from the reference frame is transmitted. .

For example, in the macro block MB (i, j), in case of DCT conversion of the field processing mode in the motion prediction of the frame processing mode, the macro block MB (i, j) is an odd number. The difference between the predicted picture extracted from the field and the original picture of the odd field, the frame motion vector of the macro block MB (i, j), and the frame motion vector of the macro block MB (i + 1, j). The macroblock MB (i + 1, j) transmits the difference between the predicted picture of the odd field and the input picture.

By the way, in the encoding device of the present embodiment, this code is realized by adding the extension bits to the conventional macroblock time and making it compatible with the conventional.

That is, in the case of the first embodiment, for example, in a B frame, there are three types of macro block types, pre-prediction, post-prediction, and both prediction, as described above. Since two methods of prediction in the odd field and the even field can be considered, the present code is realized by adding an extension bit that recognizes which prediction. Since there are two types of predictions in this case, one bit may be added to the extension bits in one direction (front and rear prediction). For example, in the case of the prediction in the odd field in the before or after prediction, the sign "1" may be added to the conventional macroblock type as the extension bit in the case of the prediction in the even field. In both predictions, both extension bits are added to the before or after prediction.

In the frame prediction mode, the extended bits are not added and have the same format as a conventional bit strip (MPEG).

The above is also applied in the case of P frames.

Next, in the case of the second embodiment, for example, in the B frame, the macroblock type includes the pre-prediction, the post-prediction, and the positive prediction, as described above. An extension bit must be added to the macroblock type to recognize whether it is prediction, prediction in the even field or prediction in the odd field in its frame. That is, in the field prediction mode of all the predictions, since there is prediction in the own frame, in order to express the three non-method predictions with the extension bits including the odd evenness, one or two bits of extension bits are required. In the field prediction mode of post-prediction, since there are only two methods of odd even, one bit is always needed for the extension. For example, in the previous prediction, in the odd field of the previous frame and in the prediction field, the sign is "1"; in the even field of the previous frame, the sign is "01"; in the odd field of the current frame, the prediction is in the odd field of the current frame. Adds the sign "11". In the post prediction, the code is "1" for prediction in the odd field of the next frame, and "0" for prediction in the even field of the next frame. This can be added to the macro block type.

In the frame mode, the extended bits are not added and are in the same format as the conventional bit stream (MPEG). In both predictions, both extension bits are added to the before or after prediction.

The same applies to the case of the P frame.

As a variant of this, it is also possible to reduce the extension bits to 1 bit in the case of the previous prediction. That is, in the even cycle in the field prediction mode, as shown in FIG. 16, all prediction is abolished by abolishing the prediction indicated by the dashed line in the figure in the odd field of the previous frame that is furthest apart temporally and positionally. It is possible to reduce to two and to transmit the full prediction mode with one bit extension. Specifically, in the previous cycle in the odd cycle, the sign is "1" for the prediction in the odd field of the previous frame, the sign "0" for the prediction in the even field of the previous frame, and in the previous prediction in the even cycle. If the prediction is in the odd field of the current frame, the code is "1"; if the prediction is in the even field of the previous frame, the code is "0"; In the case of prediction of the sign "1" and the even field of the next frame, the sign "0" may be added to the conventional macroblock type as an extension bit.

17 shows a block diagram of a decoder of an image signal. That is, the high efficiency decoding apparatus of the present embodiment receives and decodes the encoded picture data, the header information including the detected motion vector information, the motion prediction mode information, and the block mode information, and decodes the decoded picture decoded data together. The inverse variable-length encoding circuit 51 which outputs a macroblock address increment among the detected motion vector information, the motion prediction mode information, the blocking mode information, and the header information of the macroblock, and the image decoding in the macroblock address increment. The address increment values in the frame buffers 61, 62, and 64 that store data are calculated, the head addresses of the macro blocks are obtained, respectively, and the address generating carriers 81, 82, which give the head addresses to the frame buffers, respectively. 83) and the relative advice of the macro block other than the head address A data is added to the frame buffers 61, 62, and 64 to access data, receive the detected motion vector, the motion prediction mode information, and the blocking mode information, and between the motion compensation frames or fields corresponding to the mode information. It has a motion compensation circuit 59, 60, 63, 65, 66 configured to execute prediction and send motion compensated image signals to the frame buffers 61, 62, 64.

Now, in FIG. 17, the data encoded by the high efficiency encoding device of the above embodiment is once recorded on a recording medium such as a CD. The encoded data reproduced from the CD or the like is first decoded by the inverse variable-length encoding circuit 51 by the sequence, frame group, and frame by the inverse variable-length encoding circuit 51. In the odd cycle of the frame, the header portion is decoded for each slice (group of macro blocks), and the quantization width is included in the header of this slice. The macroblock address, the field processing mode / frame processing mode information, and the macroblock time indicating the decoding format are decoded for each macroblock, and the quantization width is decoded when updating.

When the blocking in the macroblock is the frame processing mode, the entire macroblock is decoded in the odd cycle and nothing is decoded in the even cycle. When the blocking is in the field processing mode, only blocks including the odd field in the macroblock are decoded in the odd cycle, and the block including the even field in the even cycle is decoded.

The image information is decoded via an inverse quantizer 53 for performing inverse quantization processing and an inverse DCT circuit 54 for performing inverse DCT conversion processing, and a macroblock type determines whether or not a differential image is obtained. In accordance with the determination result, the adder 56 switches all the switches 57 for switching between adding or not adding to the reference image (corresponding to non-intra / intra of MPEG encoding). The decoded picture is input to the frame buffer 64 or 61 (alternatively each time an I frame or P frame is processed) in the case of an I frame or a P frame, and is input to the frame buffer 62 in the case of a B frame. do. Each frame buffer is composed of two field buffers, and odd / excellent field images are stored separately in the respective field buffers. In addition, writing to the frame buffer is controlled by switching of the switch 58.

At this time, addresses written to the frame buffers 61, 62, and 64 are given by the address generators 81, 82, and 83. The address generators 81, 82, and 83 calculate the address increment values in the frame buffers 61, 62, and 64 from the macro block address increments in the header information of the macro block, and respectively, the head addresses of the macro blocks. Is seeking.

The data of both widths is stored in the I field memory 52, respectively. This quantization width data is sent to the inverse quantizer 53 via the switch 55 switched in accordance with the output of the inversely variable-length coding circuit 51. Here, in the even cycle, since only the macroblock processed in the field processing mode is decoded, the macroblock type with the address to be decoded for each macroblock and the motion vector required for the prediction method indicated by the macroblock are decoded to compensate for the motion from the reference field. The difference image that has also been transferred to the transferred image is added to obtain a reproduction.

The data of the frame buffers 64, 62, and 61 are motion compensated by the motion compensation processing circuits 65, 66, 59, 60, and 63, respectively. At this time, each motion compensation circuit switches the motion compensation / field compensation of the frame by the blocking mode (frame / field) in the DCT process.

These motion compensated images are sent to each of the selected terminals of the switching selector switches 67, 68, and 71. These switching selection switches 67, 68, and 71 are switched so that the reference field or frame indicated by the decoding method of the macroblock type can be taken out. Here, the switch selector 71 is a signal that is 1/2 of the divider 70 after the outputs of the switch selector switches 67 and 68 are added by the adder 69, and the switch 67 ) Is supplied. The output of the switch 71 is sent to the switch 57.

Moreover, the output of each frame buffer 64, 61, 62 is sent to the display 73 via the switching selection switch 72. As shown in FIG. The output of the above-mentioned switching selection switch 72 is supplied to the display 73 so that the reproduction image is displayed in order instead of the decoding order. As a result, an image is obtained.

In the above description, for example, when there is a moving body CA in front of a stationary background as shown in FIG. 22, the couple moves into a comb-type KS because there is movement between the fields when one frame is observed. According to the present embodiment, since such moving parts are encoded in the field processing mode, the moving parts can be processed as an image without field-shaking, and high-quality moving pictures with high efficiency can be obtained by compensating between odd and excellent motions. I) can be played back. That is, as shown in FIG. 18, for the odd cycle, the moving part is processed in the field processing mode while the still part is processed in the frame processing mode. In addition, the part which an image has already generate | occur | produced in an even cycle becomes the part shown by the oblique line in the figure of FIG. In the figure of FIG. 19, the moving parts other than the oblique part are decoded by the motion compensation.

In the present embodiment, however, since only the macroblock processed in the field processing mode is decoded in the even cycle, it is necessary to know the macroblock address. There are two ways of knowing the macro block address. One method is to transfer the address of the macro block for each macro block of the even cycle described above, and the other field is one field processing mode / frame processing mode in the odd cycle. Is stored in the processing mode, and the address of the macroblock in the field processing mode is converted into columns in each processing mode. The former advantage does not require the addition of memory, and the latter advantage does not increase the transmission information. Similarly, the quantization width can be realized by transferring each macroblock without taking the method of storing one field in the above-described radix cycle.

From the above description, according to the present embodiment, the processing of one frame is divided into two cycles of the radix cycle and the even cycle. In the radix cycle, the frame processing mode and the field processing mode are switched in macroblock units. Decode the field and the even field at the same time, decode only the odd field in the field process, store the quantization width in this cycle, and use the stored information in the next even cycle to compensate for motion only macroblocks in the field processing mode. Since the reproduced image can be decoded, the encoded data can be transmitted with high efficiency. In other words, it is possible to reproduce a high quality moving picture with little transmission information.

As described above, according to the high-efficiency encoding apparatus of the image signal of the present invention, field moving or frame processing can be efficiently carried out for moving pictures having a field structure, even if the moving pictures are small, the moving pictures are high, or both are mixed. Therefore, it is possible to reproduce a high quality moving picture at the time of decoding in a later high efficiency decoding apparatus of the present invention with less transmission information.

Claims (3)

A high-efficiency encoding apparatus of an image signal that encodes a macro block composed of a two-dimensional array of a plurality of pixels as a unit, the apparatus for detecting the difference between the motion vector between frames and the absolute value of each pixel in the macro block unit. Means for detecting the difference between the motion vector between the field divided by the radix or even of the scan of the frame pixel and the absolute value of each pixel in units of the macroblock, and the frame in the macroblock. By the information output from the motion detecting means, which of the frame prediction mode for performing motion compensation in units of? And the field prediction mode for performing motion compensation in units of fields in the macroblock is effective in performing motion compensation? Judging A first mode selecting means for selecting the side mode, a frame processing mode for blocking the quadrature transformation in units of the macroblock, and a field for blocking the orthogonal transformation in units of the fields in the macroblock; Second mode selecting means for determining which of the processing modes is effective in performing orthogonal transformation using information output from the motion detecting means and the first mode selecting means, and selecting an efficient blocking mode; Recognizing whether or not the odd cycle of the period in which the radix field is scanned or the even cycle of the period in which the even field is scanned in the interlacing of the encoding process for one frame. Advice to control frame memory group to output blocked macroblock Receiving the prediction generating means, the motion prediction mode information selected by the first mode selecting means, and the blocking mode information selected by the second mode selecting means, and performing motion compensation frame or inter-field prediction corresponding to the mode information. A high efficiency encoding device for an image signal, characterized by comprising motion compensation means. A high-efficiency encoding apparatus of an image signal for encoding a macroblock formed by a two-dimensional array of a plurality of pixels as a unit, comprising: means for detecting a difference between a motion vector between frames and an absolute value of each pixel in the macroblock unit; Motion detection means comprising means for detecting a difference between a motion vector between fields consisting of a radix or even number of scans of pixels of the frame in block units and an absolute value of each pixel, and motions in units of frames in the macroblock; The information output from the motion detection means determines whether the frame prediction mode for compensation and the field prediction mode for motion compensation in units of fields in the macroblock are effective in compensating for motion. Prediction mode A first mode selecting means for selecting a; and a block processing to block orthogonal transformation on the basis of the frame processing mode and the field in the macroblock to perform orthogonal transformation on a frame basis in the macroblock; Second mode selecting means for determining which of the modes is efficient in performing orthogonal transformation by using the information output from the motion detecting means and the first mode selecting means, and selecting an efficient blocking mode; 1 Recognizing whether the frame is a radix cycle in the period in which the radix field is scanned in the interlace of the encoding process or the even cycle in the period in which the even field is scanned, and the radix cycle is performed only when the blocking mode is in the field processing mode. Outputs one frame of macroblocks in Address generation means for controlling a frame memory group so as to output one even field of macroblocks sequentially in the even cycle, the motion prediction mode information selected by the first mode selection means, and the second And motion compensation means for receiving the block mode information selected by the mode selection means and performing motion compensation frame or inter-field prediction in correspondence to the mode information. Receives and decodes received picture coded data, header information including detected motion vector information, motion prediction mode information, and blocking mode information, and blocks the detected motion vector information and motion prediction mode information together with the decoded picture decoded data. An inverse variable encoding means for outputting a macroblock address increment in mode information and the header information of the macroblock, and an address increment value in a frame buffer which accumulates the image decoded data from the macroblock address increment. Calculates, obtains a head address of each macro block, adds the head generating address to the frame buffer, and adds a relative address of the macro block other than the head address to the frame buffer to access data. Receive the detected motion vector, the motion prediction mode information and the blocking mode information, perform motion compensation frame or interfield prediction corresponding to the mode information, and send a motion compensated image signal to the frame buffer. An apparatus for encoding high efficiency of an image signal, comprising: motion compensation means.