JPH08205144A - Image encoding device - Google Patents

Abstract

PURPOSE: To set different compression conditions between the inside and outside of a specified area that a user can optionally set. CONSTITUTION: The user specifies a part which is given importance of picture quality as the area through an area input means 12. An area dividing means 14 divides inputted image information into blocks of 8×8 pixels and outputs them in order to a compressing means 16. Further, the means 14 converts the area inputted through the area input means 12, block by block, and outputs binary attribute values corresponding to the respective areas to a compression condition setting means 18. The compression condition setting means 18 sets quantization tables based upon the attribute values from the means 14 in a compressing means 16 by the blocks and the compressing means 16 compresses the image blocks from the area dividing means 14 according to the quantization tables. The compressed image information is stored on a large-scale storage device or outputted to a communication line by a storing/communicating means.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image coding apparatus for compressing and coding an image.

[0002]

2. Description of the Related Art Advances in computer technology, communication technology, and image processing technology have made it possible to transfer still images, moving images, and audio even between personal computers or workstations of normal performance. Furthermore, even in a device using wireless transmission such as a portable device, a similar function is desired and is about to be put into practical use.

At present, as an image compression method, the JPEG system is standardized for still images, the MPEG system is mainly used for moving images, and the MPEG system is mainly used for storage media and ITU-T Recommendation H for communication (for video conference). ． 261 has been standardized, and many applications corresponding to these are already on the market.

These standards adopt DCT (discrete cosine transform), quantization, and entropy coding as a basic specification for an image block of 8 × 8 pixels so as to deal with images in general. Decompression of compressed data is the reverse process. The compression rate and the image quality largely depend on the values in the quantization table. In general, when a large numerical value is used for the quantization table value, the image quality deteriorates although the compression rate increases, and when a small numerical value is adopted, the image reproducibility is good but the compression rate decreases.

The JPEG system is roughly classified into sequential coding and progressive coding. In sequential coding, the final image is displayed sequentially from the upper block at the time of decoding, whereas in progressive coding, the resolution is gradually increased from the rough overall image with low resolution. Is. For image retrieval and the like, progressive coding is advantageous in that a rough image can be seen first, but in general, sequential coding that can simplify the memory configuration is used.

[0006] In the still image coding system using the DCT, when images having different characters are mixed (for example, characters and photographs), the image area is separated and the quantization is performed according to each part. Adaptive quantization using tables is known.

In image transmission, the transmission capacity is a major factor. That is, in an environment where high-speed communication can be used, the image quality can be emphasized, but in the situation where the transfer speed cannot be increased like the conventional telephone line or wireless communication, the image quality and / or the number of frames of the moving image is increased. It corresponds by dropping.

[0008]

Most of the images to be transferred usually have a mixture of important parts and non-important parts. Nevertheless, in the conventional image encoding device, the compression condition (for example, the quantization table) is set to be constant within the screen, so that the entire image has the same compression effect.

In adaptive quantization, the quantization table is changed according to the area (photograph or character), but it can only be applied to documents such as photographs and characters that can separate a specific image area. Yes, the user cannot set the area arbitrarily.

The present invention eliminates such inconvenience, and presents an image coding apparatus capable of arbitrarily designating one or more areas in an image to be coded and having different compression conditions for the designated area and other areas. The purpose is to do.

[0011]

An image coding apparatus according to the present invention includes an area setting means for setting one or more designated areas on a screen for an image to be coded, and the area setting means. Further, it is characterized by comprising an encoding means for performing image encoding under different encoding conditions for the one or more designated areas and the other areas. The area setting means is composed of an input means for inputting an area designation value indicating the designated area and a determination means for determining the designated area according to the area designation value input by the input means.

The coding condition is, for example, a quantization table value, or selection of a region band in progressive coding.

[0013]

By the area setting means, it is possible to divide into one or more designated areas and areas other than the designated areas. The encoding means encodes an image under different encoding conditions between the designated area and the other areas. Thus, the designated area arbitrarily designated by the user can be compressed at a higher compression rate or a lower compression rate than other areas.

[0014]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 shows a schematic block diagram of an embodiment of the present invention. The image information 10 to be encoded is, for example,
Input from a video camera or from a video storage device. The area input means 12 is, for example, a pointing device such as a pen or a mouse, and the user uses the area input means 12 to encode an area of the image information 10 to be encoded at a high compression rate and a low compression rate. Enter the power area.

The area dividing means 14 divides the input image information into predetermined blocks (here, 8 × 8 pixels) and outputs them to the compressing means 16 in block order. The area dividing unit 14 also outputs the binary attribute value corresponding to the area input by the area input unit 12 in synchronization with outputting the image information to the compression unit 16 in block units.
Output to 8. The area designated by the area input means 12 is in dot units or pixel units, but since the compression means 16 compresses and encodes image information in blocks of 8 × 8 pixel units, the area division means 14 will be described later. The attribute value supplied to the compression condition setting means 18 is also a block unit of the same size and position.

The compression condition setting means 18 is the area dividing means 1
The compression condition (here, the quantization table) according to the attribute value from 4 is set in the compression unit 16 in each block, and the compression unit 16 follows the set compression condition and the area dividing unit 1
Image blocks from 4 are compression encoded. The compressed image information is stored in the large-scale storage device or output to the communication line by the storage / communication means 20.

The compressed image information read from the large-scale storage device by the storage / communication means 20 or input from the communication line is expanded by the expansion means 22. The image reproduction means 24 reproduces the output of the expansion means 22, reproduces and displays it on the monitor screen, or outputs it from the printer.

FIG. 2 is a block diagram showing the internal structure of the compression means 16. Here, JPEG method or MPE
The G system I picture coding configuration is adopted. YU
The V conversion means 30 converts the image data from the area dividing means 14 into Y (luminance) and UV (color difference signals). The DCT means 32 receives the Y data and UV from the YUV conversion means 30.
By performing the discrete cosine transform on the data, the original image information of 8-dot corner is converted to the 2-dimensional frequency space of 8-dot corner.

The quantizing means 34 is a compression condition setting means 18
The output of the DCT means 32 is quantized according to the quantization table from Needless to say, quantization is performed so that the information in the high frequency region is small, preferably 0, by utilizing the fact that it is difficult to visually identify even if the information with a high spatial frequency is dropped. The quantized data is zigzag-scanned so that 0s are continuous within an 8-dot corner, and the quantized data is supplied to the Huffman encoding means 36.
The Huffman coding unit 36 assigns to the output data string of the quantization unit 34, a short code in order from the information with the highest appearance probability.

In the conventional example, there is only one quantization table used in the quantizing means 34 in one screen (frame or field), but in the present embodiment, the compression condition setting means 1 is used.
8, the different quantization table for each area designated by the area input means 12 is used within one screen. For example,
For the two types of area designation, the important image portion designated by the area is set to a small quantization table value so that the image information is not much lost, and the other areas are large enough to allow some information loss. Use the quantization table value.

Of the plurality of quantization table values used,
The smaller quantization table value is transmitted to the subsequent stage as the header of the compressed file, and is used for the decompression processing by the decompression means 22. Since only one quantization table value is used in the decompression process, the decompression means 22 may have the same configuration as the JPEG decompression means or the MPEG I-picture decompression means. Of course, depending on the transmission capacity or the degree of compression obtained, all the quantization tables used in one screen may be transmitted together with the compressed data. In that case, the decompression means 22 needs a means for switching the quantization table value to be adopted for each coding block.

The present embodiment can be realized, for example, as an electronic device whose schematic configuration is shown in FIG. The CPU 40 is a 32-bit or 16-bit controller that controls the entire device and executes applications, and the peripheral I / O 42.
Is an IC that pairs with the CPU 40 to control peripheral I / O (interrupt control, serial / parallel communication, RTC, etc.). The RAM / ROM 44 is a memory used for the control execution program and the work area, and the HDD 46.
Is a hard disk device that stores application software and user data when the power is turned off.

The display control device 48 includes a liquid crystal display panel 50.
To display the image. The digitizer 52 is used for designating an area and selecting a menu on the display screen of the liquid crystal display panel 50. It is well known that a pen locus for character input can also be input as a series of coordinate values. The voice control device 54 controls voice input / output by the microphone / speaker 56.

The image codec 58 has an input image from an attached small camera 60, an image stored in the HDD 46,
And a device for compressing / decompressing an image transmitted via communication. In the present embodiment, the MPEG system, the JPEG system, or the recommended H.264 standard is used. Any of H.261 or two or more are supported.

The communication control device 62 controls the exchange of data with other devices (such as a computer and a facsimile device) via the modem 64, a telephone line, and a LAN. The communication medium may be wired or wireless, and the data may be analog or digital.

The information device shown in FIG. 3 can be configured as a portable information device having the appearance as shown in FIG. 4, for example. The same code | symbol shows the same component. The liquid crystal display panel 50 and the digitizer 52 are superposed in a touch panel manner as is well known. As a result, if the surface of the digitizer 52 is pressed by the input pen 66 for the digitizer 52 with a predetermined pressure or more, the coordinate value of the point is input, and the locus of the writing pressure is displayed on the liquid crystal display panel 50 by known software. It The antenna 68 is equipped for wireless communication.

FIG. 5 shows an operation flowchart of this embodiment. First, on the display screen of the liquid crystal display panel 50, the digitizer 52 designates an area to be set at a low compression rate or, conversely, an area to be set at a high compression rate (S1). The designated area is converted into an area of a block unit of 8 × 8 dots for compression (S2). The details will be described later. Next, an area change point table that clearly indicates the timing of switching the compression condition during the compression work is created (S3), and an initial value is set in the pointer (S4).

A plurality of quantization tables are set on a predetermined memory (S5), the total number of blocks is set from the size of the image, and a counter indicating the current block to be processed is initialized (S6). Further, a pointer (quantization pointer) indicating the address of the quantization table used first is set (S7).

After the above initialization processing, the compression work is sequentially executed for all blocks while switching the quantization table according to the area (S8 to S13). That is, the current block counter is compared with the numerical value shown in the change point table (S8), and if they match, the block is the block for changing the quantization table, so the quantization table pointer is changed (S9). Since two quantization tables are used in this embodiment, the change in S9 means alternate switching. The pointer indicating the table indicating the change position is advanced by 1 (S10), and the compression is executed (S11).

On the other hand, when the current block counter does not match the numerical value shown in the change point table (S8), the compression is executed as it is (S11).

After the compression process (S11) is performed, the current block count is compared with the total number of blocks (S12). If they match, the process ends. If they do not match, the current block counter is incremented by 1 and S8 is performed. Repeat the above.

FIG. 6 shows an example of the quantization table in the designated area and the area outside thereof. The quantization table 1 is applied to the designated area, and the quantization table 2 is applied to general areas other than the designated area. The numerical values of the quantization tables 1 and 2 are examples, but the numerical values of the entire table are lowered in the designated area where the image quality is important, and the numerical values of the quantization table are increased in the general area where the image information is not so important. . However, in this embodiment, when decompressing, the quantization table 1 is used regardless of whether it is the designated area or the general area.
The numerical values of the low frequency region (upper left part of the table) containing the information indicating the skeleton of the image are the same.

FIG. 7 shows the relationship between the designated area and the blocks input in dot units by the digitizer 52. As described above, the compression process used in the present embodiment is executed in units of 8-dot square blocks, so that the blocks are converted in block units so that the designated area matches the blocks. The block on the boundary of the input specified area is
All are processed as a designated area.

FIG. 8 shows a detailed flowchart of the area division (S2). First, it is determined whether or not the input line segment is a closed curve (S21). If not, an auxiliary line connecting the end points is drawn to form a closed curve (S21).
22). Initialize the block counter N (S2
3).

All the blocks in the screen are sequentially examined to see if they are in the designated area. That is, it is checked whether the block indicated by the block counter N is in the designated area (S24),
If it is within the designated area (S24), the block attribute matrix H
Attribute 1 is set in (N) (S25), and attribute 0 is set outside the area (S26). If the block counter N is equal to the total number of blocks max, the process is ended. If they are not equal, N is incremented and S24 and subsequent steps are repeated (S27).

FIG. 9 shows creation of a region change point table (S3).
2 shows a detailed flowchart of An area change point table is created based on the attribute matrix H (N) created in FIG.
Initialize the pointer of the table T to be created (S3
1) The current block counter N is initialized (0 is substituted) (S32), and the attribute value H (0) of the first block is substituted into the reference attribute R (S33).

After that, the attribute change points of all the blocks are checked, and the counter N of the changed block is stored in the table T.
To be stored. That is, the reference attribute R and the current block
The attribute value H (N) of the block indicated by the counter N is compared (S34). If they do not match, the block counter N is written in the table T as a change point (S35), and the table pointer of the table T is incremented. (S
36), the reference attribute R is updated with the attribute value H (N) (S37). If there is no match in S34, or after S37, it is determined whether all blocks have been examined (S3
8) If all the blocks have been checked, the process ends, and if not checked, N is incremented (S39),
The steps from S34 onward are repeated.

FIG. 10 shows a detailed flowchart of the compression process (S11). The image data is read in block units of 8 × 8 dots (S41), and discrete cosine transform (S
42), and the transform coefficient is quantized (S43).
The quantization table used in S43 is designated by the quantization table pointer in S7 and S9. After the quantization, zigzag scanning is performed, and Huffman coding that gives shorter codes to code strings that appear more frequently is performed (S44) and stored in a predetermined storage means (S44).
45).

As the area input method by the user, the method of enclosing an image with a pen has been described. However, a rough contour is input, an object in it is automatically extracted, and the object is designated as a specified area. Good. Furthermore, in the present embodiment, the region in which the compression condition is changed is divided into blocks, but before the image outside the region is low-pass filtered to reduce the spatial frequency of the image in advance so that it becomes substantially dot units. You may add a process.

By changing the quantization table according to the area designated by the user and the area not designated by the user, the data compression rate can be appropriately distributed. Furthermore, since there is no need for special processing to switch the quantization table during decompression,
It can be played back with a normal JPEG decoder.

The above embodiment is an example of sequential encoding in which the image is sequentially compressed and stored or transmitted from the beginning, and at the time of reproduction, the image is sequentially reproduced from the upper part of the image. next,
An example of progressive coding in which information of a rough portion of an image is sequentially compressed and stored in detail information and reproduced in detail from a rough entire image during reproduction will be described.

FIG. 11 shows a block of 8 × 8 pixels after DCT. This block is further divided into a plurality of bands in the frequency band. Since band 1 has a DC component and a low frequency region, an approximate shape and color can be reproduced by reproducing only this part. That is, it is possible to determine what the image looks like, even with only band 1. The quality of the reproduced image is improved by adding the information of band 2 and thereafter. In progressive coding, quantization, Huffman coding and transmission are generally
Band 1, band 2, band 3 and band 4 are performed separately. That is, the DCT transform coefficient data of band 1 is
Band 2 after being quantized, Huffman coded and transmitted
DCT transform coefficient data is quantized, Huffman coded, and transmitted. The same applies to band 3 and band 4.

In the present embodiment, for the designated area designated by the user, all the data of bands 1 to 4 are transmitted in order to give priority to the quality of the image, and for the general area outside the designated area, the compression rate is emphasized. Then, only the data of band 1 is created and all the data of bands 2 to 4 are set to zero and transmitted.
As a result, high image quality is obtained in the designated area and high compression rate is obtained in the general area.

FIG. 12 shows an operation flowchart of this embodiment. As in the case of FIG. 5, the changed points are tabulated by the user's arbitrary area designation (S51, 5).
2, 53). DCT is executed for all blocks, and the transform coefficient data is stored in a predetermined memory (S54). For the first band, quantization and Huffman coding are executed for all blocks (S55).

Each block of all blocks is sequentially scanned, and the blocks in the designated area are also coded (quantized and Huffman coded) in the second band, the third band and the fourth band (S57, 58, 59). ). For blocks outside the designated area, the data of the second band, the third band, and the fourth band are treated as zero.

In this way, even with the progressive coding method, high image quality can be realized in the designated area and high compression rate can be realized outside the designated area, and the designated area can be arbitrarily set by the user. Depending on the desired compression ratio, the coded data of band 2 or band 2 and band 3 may be transmitted even outside the specified area.

Although the embodiment of the JPEG system has been described, the present invention is not limited to the MPEG system and Recommendation H.264. It can also be applied to a method based on H.261. That is, the MPEG system and Recommendation H.264. 26
Since the reference intra image (I picture) in 1 uses the same compression algorithm as the JPEG method, the image quality and the compression rate are changed by changing the quantization table in the area specified by the user while looking at the screen and outside the area. Can be changed. Further, by making such a change in the I picture, it also affects the predictive coding performed with the I picture as a reference, and the data is lost outside the specified area, so the change is reduced and further data compression is performed. Connect

An I picture forms a GOP (Group of Pictures) by 4 frames of P pictures and 10 frames of B pictures sandwiched between them, and these B pictures and P pictures are composed of I pictures. Data is compressed by predictive coding and motion compensation. Therefore, in order to enable further compression, the change in motion may be reflected only in the I picture outside the designated area, and may be treated as having no change in the P picture and B picture. That is, in the P picture and the B picture, the data after the designated area is set to zero. As a result, the image outside the specified area is frame-advanced for each I picture, but the normal area can be viewed in the specified area. In many cases, it is still sufficient due to the nature of the image. Of course, frame thinning such as reproducing only P pictures may be combined.

Thus, not only still images but also moving images,
The same effect can be obtained.

[0051]

As can be easily understood from the above description, according to the present invention, the compression condition can be changed inside and outside the designated area which can be arbitrarily set by the user. Accordingly, it is possible to arbitrarily provide a portion that emphasizes image quality and a portion that emphasizes compression rate, and it is possible to realize both image reproducibility and compression rate.

[0052]

[Brief description of drawings]

FIG. 1 is a schematic block diagram of a basic structure of an embodiment of the present invention.

FIG. 2 is a schematic block diagram of a compression unit 16.

FIG. 3 is a schematic configuration block diagram of hardware of a portable information device incorporating the present embodiment.

4 is an external front view of the portable information device shown in FIG.

FIG. 5 is an operation flowchart of the present embodiment.

FIG. 6 is a diagram showing a relationship between a designated area and a quantization table.

FIG. 7 is a diagram showing a relationship between an input value of a designated area and a block.

FIG. 8 is a flowchart of area division.

FIG. 9 is a flowchart for creating a change point table.

FIG. 10 is a flowchart of a compression process.

FIG. 11 is an example band of progressive coding.

FIG. 12 is an operation flowchart of an embodiment applied to progressive coding.

[Explanation of symbols]

 10: image information to be encoded 12: area input means 14: area dividing means 16: compression means 18: compression condition setting means 20: storage / communication means 22: decompression means 24: image reproduction means 30: YUV conversion means 32: DCT Means 34: Quantization means 36: Huffman coding means 40: CPU 42: Peripheral I / O 44: RAM / ROM 46: Hard disk device 48: Display control device 50: Liquid crystal display panel 52: Digitizer 54: Voice control device 56: Microphone / speaker 58: Image codec 60: Small camera 62: Communication control device 64: Modem 66: Input pen 68: Antenna

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical indication H03M 7/40 9382-5K H04N 1/413 D

Claims (4)

[Claims] 1. For an image to be encoded, 1 in the screen
Area setting means for setting the above-mentioned designated area, and coding means for image-coding the one or more designated areas set by the area setting means and the other areas under different coding conditions are provided. An image encoding device characterized by the above. 2. The area setting means comprises input means for inputting an area designation value indicating the designated area, and determining means for determining the designated area according to the area designation value input by the input means. The image coding apparatus according to claim 1. 3. The image coding apparatus according to claim 1, wherein the coding condition is a quantization table value. 4. The image coding apparatus according to claim 1, wherein the coding condition is selection of a region band.