JPH07162689A - Picture encoding device - Google Patents

Abstract

PURPOSE:To shorten processing time by detecting a distortion amount, selecting data irreversibly encoded in a first encoding part when the distortion amount is equal to or less than a distortion amount threshold value and selecting reversibly encoded data from a second encoding part when it exceeds the threshold value. CONSTITUTION:Irreversible encoding is performed for respective blocks in an orthogonal transformation encoding part 12. Quantization is performed in a quantization part 15, division into planes is performed for respective bits in a part for turning the bits to the planes and reversible encoding is performed by using arithmetic encoding for respective bit planes in a binary picture encoding part 17. Selection information from a distortion calculation part 14 is supplied to a selection means 18. The means 18 selects the encoded data from the encoding part 12 and outputs them to a multiplexing part 19 by the selection information when the calculated distortion amount is equal to or less than the distortion amount threshold value. Also, the encoded data from the encoding part 17 are selected and outputted to the multiplexing part 19 by the selection information at the time of exceeding the distortion amount threshold value. Thus, since both orthogonal transformation encoding and arithmetic encoding are simultaneously performed, the processing time is shortened.

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 coding an image signal.

[0002]

2. Description of the Related Art When an orthogonal transform coding method represented by the international standard color still picture coding method is applied to an image in which characters and photographs are mixed, the image quality of characters in a reproduced image obtained by decoding remarkably deteriorates. The problem of becoming is known. This is because the lossy encoding method that realizes data compression by deleting the high frequency component of the image signal is adopted, and the reproducibility of a portion having a high frequency component such as an edge of a character is deteriorated.

In order to solve this problem, Japanese Patent Laid-Open No. 4-35
As disclosed in Japanese Patent No. 6873, the input image is separated into a character image and a natural image by image area separation processing, and the character image is
There is a method in which a binary image is converted by converting it into a value image and a natural image is subjected to natural image coding such as orthogonal transform coding.

The method will be described in more detail as follows.
As shown in FIG. 2, the image data output from the scanner 1 that reads an image is output in the order of raster scanning, and is temporarily stored in a buffer 2 that stores image data for several lines. Dividing part 3
After dividing into blocks, it is determined in block units whether the area is a character / background area or a picture area. This determination is performed by the image area separation unit 4. That is, the image area separating unit 4 obtains the number of white pixels in the block, and if there are more white pixels than the threshold value and no color pixels, it is set as a white background block. To do. If there is an edge area in the block, the area is a character area, and areas other than the character area and the background area are pattern areas.

After performing the above-described image area separation processing in the image area separation unit 4, the image data of the block judged as the character / background area is input to the binary image encoding unit 6 by controlling the switch unit 5. After converting to a binary signal, lossless encoding suitable for the binary image is performed and output to the multiplexing unit 7, and the image data of the block determined to be the picture area is controlled by the switch unit 5 to the natural image encoding unit. It is input to 8 and encoded suitable for a natural image and output to the multiplexing section 7.

The area information coding section 9 codes the information on the area determined by the image area separating section 4 as area information and outputs it to the multiplexing section 7.

The multiplexing section 7 multiplexes the coded data from the binary image coding section 6 or the natural image coding section 8 and the coded data from the area information coding section 9 and sends it as coded data. It is like this.

[0008]

However, in the above-mentioned publication, since image area separation is performed as preprocessing for encoding processing, hardware for image area separation processing is added and processing time increases. I have a problem.

Further, since the character / background area is converted from a gradation signal into a binary signal and coded as a binary image, for example, although it is actually a picture area, some blocks of the area are in the character / background area. If the area is erroneously determined, the quality of the picture area in the reproduced image is significantly deteriorated due to the erroneously determined block. Therefore, there is a problem that the determination accuracy of the image area separation processing must be made extremely high.

Therefore, according to the present invention, an image in which characters and photographs are mixed can be appropriately coded without performing image area separation processing, and therefore the quality of a reproduced image can be maintained, and the processing time can be shortened. An image encoding device capable of performing the above is provided.

[0011]

According to the present invention, there is provided a block division section for dividing an image signal into blocks composed of a plurality of pixels, and lossy encoding of image data of each block divided by the block division section. No. 1 encoding unit, a local decoding unit that locally decodes the data lossy-encoded by the first encoding unit to obtain reproduced image data, and the image data of each block divided by the block dividing unit is quantized. A quantization unit that
A bit plane converting unit that divides the image data quantized by the quantizing unit into planes for each bit, and a second encoding unit that performs lossless encoding for each bit plane divided by the bit plane converting unit, The distortion amount between the reproduced image data from the local decoding unit and the original image data from the block division unit is calculated, and it is determined whether or not the calculated distortion amount is equal to or less than a preset distortion amount threshold value and the selection information is output. Based on the distortion calculation unit and the selection information of the distortion calculation unit, when the distortion amount is equal to or less than the distortion amount threshold value, the data lossy-encoded by the first encoding unit is selected and output, and the distortion amount is the distortion amount threshold value. And a selecting means for selecting and outputting the reversible coded data from the second coding section.

[0012]

According to the present invention having such a configuration, the image data of each block divided by the block dividing unit is lossy-encoded by the first encoding unit, and also quantized by the quantizing unit, and the bit-plane converting unit is quantized. Then, the data is divided into planes for each bit, and then the second encoding unit performs lossless encoding. Then, the distortion amount between the reproduced image data obtained by locally decoding the data lossy-encoded by the first encoding unit by the local decoding unit and the original image data from the block dividing unit is calculated, and the distortion amount is calculated based on the calculation result. Is less than or equal to the distortion amount threshold value, the data lossy-encoded by the first encoding unit is selected and output, and when the distortion amount exceeds the distortion amount threshold value, the lossless encoded data from the second encoding unit is selected. And output.

[0013]

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

As shown in FIG. 1, the image signal is divided into blocks of, for example, 8 × 8 pixels by the block dividing unit 11. Then, the orthogonal transform coding unit 12, which is the first coding unit, performs orthogonal transform coding represented by the color still image international standard coding method on each of the divided blocks. That is, lossy encoding is performed.

The coded data obtained by the orthogonal transform coding in the orthogonal transform coding unit 12 is locally decoded in the local decoding unit 13 to obtain reproduced image data of the block, and the reproduced image data is supplied to the distortion calculation unit 14. is doing.

The distortion calculator 14 calculates the amount of distortion from the original image data of the block before input to the orthogonal transform encoder 12 and the reproduced image data of the corresponding block from the local decoder 13.

The distortion calculation is obtained, for example, according to the following equation.

[0018]

[Equation 1] Here, ΔX is the distortion amount, X _ij is the original image data of the 8 × 8 pixel block, and Y _ij is the reproduced image data of the 8 × 8 pixel block.

Then, it is determined whether or not the calculated strain amount is less than or equal to a preset strain amount threshold value, and the selection information based on the determination result is output.

Further, the block image data from the block dividing unit 11 is quantized by the quantizing unit 15 at 2 bits / pixel or more. For example, if the block image data is 8 bits / pixel, it is quantized into a multi-level signal of 4 bits / pixel. Then, the quantized image data is transferred to the bit plane conversion unit 1
Supply to 6.

The bit plane conversion section 16 divides the quantized image data into planes for each bit. For example, if the quantized image data is 4 bits / pixel, it is divided into 4 bit planes. Then, the bit planes divided by the binary image encoding unit 17, which is the second encoding unit, are losslessly encoded for each bit plane. For example, the coding is performed using arithmetic coding represented by the international standard binary image coding method.

The selection information from the distortion calculation section 14 is supplied to the selection means 18. The selecting unit 18 selects the encoded data from the orthogonal transform encoding unit 12 according to the selection information when the calculated distortion amount is less than or equal to the distortion amount threshold value, outputs the coded data to the multiplexing unit 19, and calculates the calculated distortion amount. The coded data from the binary image coding unit 17 is selected according to the selection information when the amount exceeds the distortion amount threshold value, and is output to the multiplexing unit 19.

The selection information from the distortion calculation section 14 is supplied to the selection information coding section 20. The selection information encoding unit 20 encodes the selection information of, for example, 1 bit / block from the distortion calculation unit 14 by arithmetic encoding represented by, for example, a binary image international standard encoding method, and then the multiplexing unit. Supply to 19.

When the coded data from the orthogonal transform coding unit 12 is selected, the multiplexing unit 19 codes the coded data from the selection information coding unit 20 and the coded data from the orthogonal transform coding unit 12. The data is multiplexed and transmitted as encoded data, and when the encoded data from the binary image encoding unit 17 is selected, the encoded data from the selection information encoding unit 20 and the binary image encoding unit are selected. The coded data from 17 is multiplexed and sent as coded data.

In the embodiment having such a configuration, the image data of each block divided by the block division unit 11 is orthogonally transformed and coded by the orthogonal transformation coding unit 12. Then, the coded data subjected to the orthogonal transform coding is locally decoded by the local decoding unit 13 and reproduced.

The image data for each block divided by the block dividing unit 11 is quantized by a quantizing unit 15 into a multi-valued signal of 4 bits / pixel, and then the bit plane converting unit 1
It is divided into planes for each bit at 6, and is further arithmetically encoded for each bit plane by the binary image encoding unit 17.

In the distortion calculator 14, the original image data of the block before being input to the orthogonal transform encoder 12 and the local decoder 13 are input.
The distortion amount is calculated from the reproduced image data of the corresponding block from, and the distortion amount is compared with the distortion amount threshold value. The selection information for selecting the encoded data is transmitted, and if the calculated distortion amount exceeds the distortion amount threshold value, the selection information for transmitting the encoded data from the binary image encoding unit 17 is transmitted to the selection unit 18.

In this way, the selecting means, when the distortion amount calculated by the distortion calculating unit 14 is less than or equal to the distortion amount threshold value, the orthogonal transform encoding unit 1
The encoded data from the binary image encoding unit 17 is selected and output to the multiplexing unit 19, and when the amount of distortion calculated by the distortion calculating unit 14 exceeds the distortion amount threshold value, the encoded data from the binary image encoding unit 17 is selected. And outputs it to the multiplexing unit 19.

The coded data thus orthogonally coded by the orthogonal transform coding unit 12 is locally decoded by the local decoding unit 13, and the distortion calculation unit 14 inputs the data of the block before input to the orthogonal transform coding unit 12. The distortion amount is calculated from the original image data and the reproduced image data of the corresponding block from the local decoding unit 13, and the distortion amount is compared with the distortion amount threshold value, so that the selecting unit 18 causes the encoded data from the orthogonal transform encoding unit 12 to be selected. Is selected or the coded data from the binary image coding unit 17 is selected, preprocessing such as the conventional image area separation processing can be eliminated, and orthogonal transform coding and arithmetic are performed. Two encodings can be performed at the same time, and the processing time can be shortened.

Regarding the distortion amount calculated by the distortion calculation unit 14, the block whose distortion amount exceeds the distortion amount threshold is a block having many high frequency components such as characters, and this block is reversible as a multi-valued signal of about 4 bits. Since the data is encoded and output, there is no quality deterioration due to encoding, and therefore the reproduction quality of characters and the like can be maintained without deleting high-frequency components.

By the way, when the present apparatus is installed in a facsimile or the like, it usually has a receiving function as well as a transmitting function, so a decoding apparatus is inevitably installed. Therefore, the local decoding unit used in this device can also serve as the decoding device, and in such a case, it is not necessary to separately provide the local decoding unit, and the hardware can be reduced.

Further, since the binary image coding unit 17 and the selection information coding unit 20 use a common arithmetic coding system, the binary image coding unit 17 and the selection information coding unit 20. Since 20 can be shared by one encoding unit, it is possible to reduce hardware in this respect as well.

[0033]

As described above, according to the present invention, an image in which characters and photographs are mixed can be appropriately coded without performing image area separation processing, and therefore the quality of a reproduced image can be maintained and the processing time can be improved. It is possible to provide an image coding apparatus capable of shortening the length.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a block diagram showing a conventional example.

[Explanation of symbols]

 11 ... Block division unit 12 ... Orthogonal transform coding unit (first coding unit) 13 ... Local decoding unit 14 ... Distortion calculation unit 15 ... Quantization unit 16 ... Bit plane conversion unit 17 ... Binary image coding unit ( Second encoding unit) 18 ... Selection means

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical indication H04N 1/415

Claims (1)

[Claims] 1. A block division unit for dividing an image signal into blocks composed of a plurality of pixels, a first encoding unit for irreversibly encoding image data of each block divided by the block division unit, and A local decoding unit for locally decoding the data lossy-encoded by the first encoding unit to obtain reproduced image data; a quantizing unit for quantizing the image data of each block divided by the block dividing unit; A bit plane converting unit that divides the image data quantized by the quantizing unit into planes for each bit; a second encoding unit that losslessly encodes each bit plane divided by the bit plane converting unit; The distortion amount between the reproduced image data from the local decoding unit and the original image data from the block dividing unit is calculated, and the selection information is output by determining whether the calculated distortion amount is equal to or less than a preset distortion amount threshold value. Do Based on the distortion calculation unit and the selection information of the distortion calculation unit, when the distortion amount is equal to or less than the distortion amount threshold value, the data lossy-encoded by the first encoding unit is selected and output, and the distortion amount is the distortion amount. An image coding apparatus, comprising: selecting means for selecting and outputting the lossless coded data from the second coding section when the threshold value is exceeded.