JPH08167030A - Device and method for image compression - Google Patents

Abstract

PURPOSE: To make possible compression processing which is good in efficiency and minimizes deterioration with compression by properly demarcating reversible compression processing and irreversible compression processing. CONSTITUTION: An irreversible compression part 203 outputs irreversibly compressed data and a reversible compression part 204 outputs reversibly compressed data. A counter 206 counts the cumulative information amount of the reversibly compressed data and a comparator 210 outputs an L-level signal when the counted value is smaller than the upper-limit information amount of a block which is currently processed. Further, a color number decision part 101 detects the number of significant pixels in the block and a comparator 103 outputs an L-level signal to an OR circuit 104 when a specific value is not exceeded. A selector 121 selects the reversibly compressed data when a select signal is at L level or the irreversibly compressed data at H level and stores it in a memory 213.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image compression apparatus and method thereof, and more particularly to an image compression method and apparatus for compressing a multi-valued image and storing it in a memory.

[0002]

2. Description of the Related Art As a standard image compression method for compressing still images, JPEG (Joint Photographic Expert Group
There is a baseline system proposed in p). This is a method of performing compression with high efficiency while making the deterioration due to compression inconspicuous in consideration of human visual characteristics, and using irreversible coding. Therefore, when an image including characters, graphics, etc. is compressed, distortion called mosquito noise is generated at the edge portion, which deteriorates the image quality.

In order to solve the above problems, conventionally,
An image compression that includes a lossy compression unit based on a JPEG baseline system and a lossless compression unit that does not cause deterioration, and outputs either encoding result as final compressed data depending on the amount of information generated by both. There was a formula.

The above-mentioned conventional image compression method will be described with reference to FIG. In FIG. 5, 201 is an image input terminal, 202 is a blocking unit, 203 is a lossy compression unit, 2
Reference numeral 04 is a lossless compression unit, 205 is a threshold value input terminal, 206
Is a counter, 207 is a multiplier, 208 is a block count unit, 209 is an adder, 210 is a comparator, 211 is a subtractor, 212 is a selector, and 213 is a memory. First, image data is input from the image input terminal 201. As the image data, for example, 8-bit data of each color of NTSC-RGB is input in a raster form. Then, the input image data is divided into blocks by the blocking unit 202 in units of image pressure processing. The block at this time is, for example, J
It is composed of 8 × 8 pixels which is a compression unit such as PEG. The block image data is the lossy compression unit 203.
Is input to the lossless compression unit 204, each block is subjected to compression processing, and each compression result is output to the A terminal and the B terminal of the selector 212.

Further, the lossless compression unit 204 outputs the amount of generated information (Cn: n is the number of blocks) generated when the lossless compression processing is performed, and the counter 206 accumulates and adds. This accumulated generated information amount is ΣCn. Further, the block counting unit 208 counts the number n of processing blocks up to the present. From the threshold input terminal 205, the amount of information assigned per block is input as the threshold TH1. This threshold value TH1 is calculated by back-calculating the amount of information assigned to one block from the capacity of the memory 213. Next, the multiplier 207 multiplies the threshold value TH1 by the block number n, and calculates the sum of the information amounts allocated up to the block number n as TH1 × n. next,
TH1 × n and the surplus information amount S are input to the adder 209 and added. The surplus information amount S is a value obtained by subtracting the cumulative generated information amount ΣCn, which is the sum of the actually generated information amounts, from the sum TH1 × n of the information amounts allocated up to the current processing block, which is known from the capacity of the memory 212. . The initial value of the surplus information amount S is "0". And adder 2
The value of the output TH1 × n + S of 09 is the upper limit information amount up to the current processing block. The upper limit information amount and the cumulative generated information amount ΣCn are input to the comparator 210 and compared.

The comparison result of the comparator 210 is input as a selection signal of the selector 212. Incidentally, the comparator 210
According to the comparison result, the output from is at L level when TH1 × n + S ≧ ΣCn and at H level when TH1 × n + S <ΣCn. That is, the L level signal is output when the total amount of actually generated information is less than or equal to the upper limit information amount, and conversely, the H level signal is output when the total amount of information amount exceeds the upper limit information amount. Further, the subtracter 211 takes the difference between the upper limit information amount TH × n + S and the cumulative generated information amount ΣCn when the comparison result is the L level, and sets the result as a new surplus information amount S ′ in the adder 20.
Enter in 9. On the other hand, when the comparison result is H level,
The output of the subtracter 211 holds and outputs the previous surplus information amount S as it is.

In the selector 212, the compression result of the lossless compression unit 203 and the lossy compression unit 204 are based on the comparison result.
And one of the compression results of. That is, if the comparison result is the L level, the compression result from the lossless compression unit 204 on the B terminal side is displayed, and if the comparison result is the H level, the lossy compression unit 2 on the A terminal side is displayed.
Select the compression result from 03. Then, the selected compressed data is stored in the memory 213.

Since the lossy compression unit 203 must of course compress the data so that the amount of information is less than or equal to the upper limit information,
The output TH1 × n + S of the adder 209 is input to the lossy compression unit 203. In the lossy compression unit 203,
The compression parameter of itself is changed so that the output compressed information amount is equal to or less than the input upper limit information amount. Although details of changing the compression parameter are not described in detail, for example, JP
In the EG baseline system, it is possible to adjust the compression rate by changing the quantization step during quantization.

As described above, the input image data is compressed by the lossless compression unit, and the data compressed by the lossless compression unit is output or compressed by the lossy compression unit according to the accumulated value of the generated information amount. I decided whether to output the data. Therefore, the compressed data could be efficiently stored in the limited memory capacity.

[0010]

However, when the lossless compression process and the lossy compression process are separated by using the generated information amount as shown in the above-mentioned conventional example, the lossy compression process may be performed depending on the image to be compressed. There has been a case where reversible compression processing is performed on a region to be performed. Therefore, as a result, the memory is wasted, and the lossy compression process is performed because a sufficient free memory area cannot be secured for the part where the lossless compression process should be originally performed, resulting in deterioration due to compression. It may happen.

An example in which the above-mentioned inconvenience occurs will be described below with reference to FIG.

FIG. 6 shows an example of an image in which a natural image area and a character area are mixed in one page. For this image, the compression processing shown in the above-described conventional example is sequentially performed from the upper end of the image. At the beginning, in order to compress the character area where the amount of information is small even in lossless compression processing,
The losslessly compressed data will be selected. When the amount of information generated at this time is smaller than the amount of information allocated to one block, a surplus information amount is generated and is gradually accumulated.

Even if the processing extends to the natural image area, the amount of surplus information is accumulated, so that the reversibly compressed data is selected. Therefore, the accumulated amount of information is exhausted by compressing the natural image area to some extent.
After that, additional decompression is performed.

Further, when compressing the next character area,
In the lossless compression process, a relatively large amount of information is generated, and since the surplus information amount is not accumulated, there is no choice but to perform lossy compression. Therefore, deterioration due to mosquito noise occurs in the character area.

Originally, in the case of an image as shown in FIG. 6, the surplus information amount generated in the compression processing of the first character area is not used in the natural image area, but is allocated to the subsequent character area for lossless compression. You should do it. This is because generally in a natural image area, deterioration due to compression is hard to be visually detected even if lossy compression is performed.

As described above, in the image compression processing as described in the conventional example, it is difficult to perform appropriate compression processing on each area for an image in which characters and natural images are mixed. It was

The present invention has been made in order to solve the above-mentioned problems, and it is efficient by appropriately separating lossless compression processing and lossy compression processing, and further, deterioration due to compression is minimized. It is an object of the present invention to provide an image compression apparatus and method capable of performing the above compression processing.

[0018]

[Means for Solving the Problems] As one means for achieving the above object, the present invention has the following configuration.

That is, blocking means for dividing the input image into blocks, lossless compression means for performing lossless compression processing in block units, lossy compression means for performing lossy compression processing in block units, and accumulation by the lossless compression means. It is characterized by further comprising selection means for selecting one of the compression data by the lossless compression means and the compression data by the irreversible compression means in the block unit based on the generated information amount and the number of colors in the block.

For example, the selecting means counts means for counting the accumulated generated information amount, calculating means for calculating an upper limit generated information amount allowed up to the currently processed block, and detection for detecting the number of colors. Means for selecting the compressed data by the reversible compression means when the cumulative generated information amount is less than or equal to the upper limit generated information amount and the number of colors is less than or equal to a predetermined value, and in other cases. The compressed data by the lossy compression means is selected.

For example, the lossy compression means is characterized by performing lossy compression processing based on the upper limit generated information amount.

For example, the lossless compression means and the lossy compression means are characterized by simultaneously compressing the same block.

Further, according to another preferred embodiment, a blocking means for blocking an input image into blocks, a lossless compression means for performing a lossless compression process in the block units, and data in a frequency space in the block units. Lossy compression means for converting and performing lossy compression processing; compressed data by the lossless compression means and compression by the lossy compression means based on the amount of accumulated information generated by the lossless compression means and the high frequency power value in the frequency space It is characterized by having a selecting means for selecting any of the data in block units.

For example, the irreversible compression means may include a conversion means for converting the image data into a frequency space and a detection means for scanning the data in the frequency space to detect the maximum value of the high frequency power. Characterize.

For example, the selecting means has a counting means for counting the accumulated occurrence information amount and a calculating means for calculating an upper limit occurrence information amount allowed up to the currently processed block, If the amount is less than or equal to the upper limit generated information amount and the maximum value of the high frequency power is greater than or equal to a predetermined value, the compression data by the lossless compression unit is selected, and in other cases, the compression by the lossy compression unit is performed. Characterized by selecting data.

For example, the irreversible compression means is characterized by performing irreversible compression processing based on the upper limit generated information amount.

For example, the lossless compression means and the lossy compression means are characterized by simultaneously compressing the same block.

Further, the present invention is characterized by further comprising holding means for holding the compressed data selected by the selecting means.

[0029]

With the above configuration, it is possible to distinguish between lossless compression processing and lossy compression processing based on the number of colors in the block and the value of the amount of information generated in the lossless compression unit.

Further, a peculiar function that the lossless compression process and the lossy compression process can be distinguished based on the high frequency power value in the frequency space of the block and the value of the information amount generated in the lossless compression unit. The effect is obtained.

[0031]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment according to the present invention will be described below in detail.

<First Embodiment> FIG. 1 shows a block configuration of a compression processing unit according to the present embodiment. In FIG. 1, the same numbers are given to the same configurations as those of FIG. 5 showing the above-mentioned conventional example,
Description is omitted.

The configuration shown in FIG. 1 is obtained by adding a color number determination unit 101, a comparator 103, and an OR circuit 104 to the configuration of FIG. 5 described above, and 102 is a threshold value T.
This is an input terminal to which H2 is input.

First, the data blocked by the blocking unit 202 is input to the color number determination unit 101. The color number determination unit 101 determines the number of colors in the block. For example,
When the block is composed of 8 × 8 64 pixels, the value of the color number output from the color number determination unit 101 is “1” to “6”.
4 ”. A threshold TH2 indicating the upper limit of the number of colors is input from the input terminal 102 and input to the comparator 103.
The comparator 103 compares the color number of the current block output from the color number determination unit 101 with the threshold value TH2. The output from the comparator 210 is L level when N ≦ TH2 and H level when N> TH2 according to the comparison result, where N is the number of colors. That is, the L level signal is output when the number of colors in the block is less than or equal to the upper limit, and conversely, the H level signal is output when the number of colors exceeds the upper limit.

The comparison result from the comparator 103 is input to the OR circuit 104 and is compared with the comparison result from the comparator 210.
Taking R, the result becomes the selection signal of the selector 212. In the selector 212, the comparator 210 and the comparator 103
Only when both outputs are at the L level, that is, when the amount of information generated up to the current block is less than or equal to the upper limit information amount in the block and the number of colors in the block is less than or equal to the upper limit color number. Only, the selector 212 has the B terminal side,
That is, the lossless compression unit 204 side is selected.

In general, the number of colors in one block tends to increase in a natural image. Therefore, in this embodiment, by appropriately setting the threshold value TH2, the natural image region and the character region can be divided to some extent. Can be separated.

As described above, according to this embodiment, not only the amount of information after compression but also the number of colors in a block can be used to distinguish between a natural image area and a character area. Therefore, it is possible to perform appropriate compression processing on each area and to effectively use the memory while minimizing deterioration due to compression.

<Second Embodiment> A second embodiment according to the present invention will be described below.

The block configuration of the compression processing section of the second embodiment is shown in FIG. In FIG. 2, FIG. 5 showing the above-mentioned conventional example.
The same components as those in FIG.

The configuration shown in FIG. 2 is different from the configuration of FIG. 5 described above in that the lossy compression section 401 sends region information to the signal line 402.
It is characterized in that it is inputted to the OR circuit 403 via.

The irreversible compression section 401 uses a JPEG baseline system in order to compress a natural image or the like with high efficiency. In the JPEG method, DCT (Discrete Cosin)
e Transform) is used to transform image data into a frequency space, and coarse quantization is performed on high-frequency components that are inconspicuous due to human visual characteristics, thereby reducing the amount of information. Therefore, it is possible to divide the image area by using the data in the frequency space converted by the DCT.

Generally, in a natural image, electric power tends to be concentrated in the low frequency part in the frequency space. In addition, with regard to characters and graphic images, high power is generated up to a high frequency part in the frequency space. Therefore, in the second embodiment, the lossy compression unit 401 separates a natural image from a character image by utilizing the property in frequency space, and outputs the separation information from the signal line 402.

The lossy compression section 4 in the second embodiment will be described below.
The processing in 01 will be described. FIG. 3 shows a block configuration of the lossy compression unit 401.

In FIG. 3, reference numeral 501 denotes a DCT section, 502
Is a zigzag converter, 503 is a high frequency power MAX value detector,
Reference numeral 505 is a quantization unit, 506 is a comparator, and 507 is a VLC unit. First, the DCT unit 501 performs D
CT processing is performed and it is converted into data in the frequency space. Next, the zigzag conversion unit 502 converts the scan direction so that each data in the block is arranged one-dimensionally from the low frequency component to the high frequency component. This scanning method is shown in FIG.
As shown in FIG. 4, the zigzag conversion unit 502 performs so-called zigzag scanning in which AC components are sequentially scanned from low-frequency components to high-frequency components except for the DC component at the upper left of the block.

Subsequently, the frequency-converted data is quantized by the quantizer 505, and the VLC unit 507 performs compression processing by variable length coding. The details of these processes will not be described in detail.

On the other hand, the one-dimensional array data in the zigzag conversion section 502 is input to the high frequency power MAX value detection section 503. The high frequency power MAX value detection unit 503 detects the maximum component value among the components higher in frequency than the predetermined frequency in the frequency space, and outputs the detected value as the MAX value. For example, in the block shown in FIG. 4, from a component in a higher frequency range than a component marked with a circle, that is, a component arranged behind the component marked with a zigzag scan,
Detect the MAX value.

The MAX value is input to the comparator 506, and at the same time, the threshold value TH3, which is the upper limit of the power value generated in the high frequency range, is input from the input terminal 504. MA in the comparator 506
The X value is compared with the threshold value TH3. The output from the comparator 506 is that when the MAX value output from the high frequency power MAX value detecting unit 503 is M, according to the comparison result, when M ≧ TH3, L level, when M <TH3, H level is output. It That is, when the MAX value is larger than the threshold value TH3, it is considered that there is a component having large power in the high frequency band, and it is determined that the block is likely to be a character portion, and an L level signal is output. . On the contrary, when the MAX value is smaller than the threshold value TH3, it is considered that there is not a high-power component in the high frequency region, and it is determined that the block is likely to be a natural image portion, and an H level signal is output. To do.

The comparison result in the comparator 506 is output from the signal line 402 shown in FIG. 2 and input to the OR circuit 403. In the OR circuit 403, the value of the signal line 402 and the output result of the comparator 210 are ORed and input to the selector 212. In the selector 212, when both the value of the signal line 402 and the output result of the comparator 210 are L level, that is, the amount of information generated up to the block is less than or equal to the upper limit information amount in the block, and The selector selects the B terminal side, that is, the reversible compression unit 204 side only when a high power value exists in the high frequency part where the frequency of the block is frequency-converted. Therefore, the natural image and the character area can be distinguished more strictly.

As described above, according to the second embodiment,
Not only the amount of generated information but also the data in the frequency space of the block can be referred to separate the natural image from the character area, and the optimum compression processing can be performed for each.

Further, considering an image area in which characters are combined on a natural image, large power is generated in a high frequency region in a frequency space in a block in which the edges of characters are conspicuous. Therefore, if lossy compression processing is performed on such a block, deterioration such as mosquito noise will occur at the edge portion. However, according to the third embodiment, since the MAX value of the high frequency power in the frequency space becomes large in such an image, when the amount of surplus information has a margin, the output result of the lossless compression processing is performed. Therefore, it is possible to suppress the occurrence of image quality deterioration due to compression.

The present invention can be applied to any device such as a printer, a facsimile device, a copying machine, etc. as long as it is an image processing device for compressing an image.

The present invention may be applied to a system composed of a plurality of devices or an apparatus composed of one device. Further, it goes without saying that the present invention can be applied to the case where it is achieved by supplying a program to a system or an apparatus.

[0053]

As described above, according to the first aspect of the present invention, the lossless compression process and the lossy compression process are performed based on the number of colors in the block and the value of the information amount generated in the lossless compression unit. By separating the processing from the processing, it is possible to perform the compression processing with the deterioration of the compression minimized while efficiently using the memory.

According to the second aspect of the present invention,
Based on the high frequency power value in the frequency space of the block and the value of the amount of information generated in the lossless compression unit, the lossless compression process and the lossy compression process are distinguished from each other. It is possible to perform compression processing with minimal deterioration. Furthermore, even in an image area in which characters are combined on a natural image, lossless compression processing can be performed if the surplus information amount has a margin, so deterioration due to compression can be suppressed.

[0055]

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a compression processing unit according to an embodiment of the present invention.

FIG. 2 is a block diagram showing the configuration of a compression processing unit according to a second embodiment of the present invention.

FIG. 3 is a block diagram showing a configuration of a lossy compression unit in the second embodiment.

FIG. 4 is a diagram for explaining zigzag scanning in the second embodiment.

FIG. 5 is a block diagram showing a configuration of a conventional compression processing unit.

FIG. 6 is a diagram showing an example of an input image that is deteriorated by a conventional compression process.

[Explanation of symbols]

 101 color number determination unit 202 blocking unit 203, 401 lossy compression unit 204 lossless compression unit 212 selector 213 memory 503 high frequency power MAX value detection unit

Claims (18)

[Claims] 1. Blocking means for dividing an input image into blocks, lossless compression means for performing lossless compression processing in block units, lossy compression means for performing lossy compression processing in block units, and accumulation by the lossless compression means. An image characterized by comprising selection means for selecting either the compressed data by the lossless compression means or the compressed data by the irreversible compression means in block units based on the amount of generated information and the number of colors in the block. Compressor. 2. The selecting means includes a counting means for counting the cumulative generated information amount, a calculating means for calculating an upper limit generated information amount allowed up to a currently processed block, and a detection for detecting the number of colors. And a means, the cumulative generated information amount is less than or equal to the upper limit generated information amount,
The image according to claim 1, wherein the compressed data by the lossless compression means is selected when the number of colors is a predetermined value or less, and the compressed data by the lossy compression means is selected in other cases. Compressor. 3. The image compression apparatus according to claim 2, wherein the lossy compression unit performs lossy compression processing based on the upper limit generated information amount. 4. The image compression apparatus according to claim 1, wherein the lossless compression unit and the lossy compression unit simultaneously compress the same block. 5. The image compression apparatus according to claim 1, further comprising a holding unit that holds the compressed data selected by the selecting unit. 6. Blocking means for dividing an input image into blocks, lossless compression means for performing lossless compression processing in block units, and lossy compression for performing lossy compression processing by converting data in frequency space in block units. Means for selecting the compression data by the reversible compression means or the compression data by the lossy compression means on the basis of the cumulative generated information amount by the lossless compression means and the high frequency power value on the frequency space An image compression apparatus comprising: 7. The lossy compression means includes a conversion means for converting image data into a frequency space, and a detection means for scanning the data in the frequency space to detect a maximum value of high frequency power. The image compression apparatus according to claim 6, which is characterized in that. 8. The cumulative selection information includes: counting means for counting the cumulative generated information quantity; and calculating means for calculating an upper limit generated information quantity allowed up to a currently processed block. The amount is less than or equal to the upper limit generated information amount,
The compressed data by the lossless compression means is selected when the maximum value of the high frequency power is a predetermined value or more, and the compressed data by the lossy compression means is selected in other cases. 7. The image compression device according to 7. 9. The image compression apparatus according to claim 8, wherein the lossy compression means performs lossy compression processing based on the upper limit generated information amount. 10. The image compression apparatus according to claim 6, wherein the lossless compression unit and the lossy compression unit simultaneously compress the same block. 11. The image compression apparatus according to claim 6, further comprising holding means for holding the compressed data selected by the selecting means. 12. An input image is subjected to lossless compression processing and lossy compression processing in block units to generate lossless compressed data and lossy compressed data, respectively, and the accumulated generated information amount of the lossless compressed data and the number of colors in the block. An image compression method, characterized in that either the lossless compression data or the lossy compression data is selected in block units based on the above. 13. A method of calculating an upper limit occurrence information amount allowed up to the currently processed block, and the cumulative occurrence information amount is equal to or less than the upper limit occurrence information amount and the color number is equal to or less than a predetermined value. 13. The image compression method according to claim 12, wherein the lossless compression data is selected as the first lossless compression data, and the lossy compression data is selected in other cases. 14. The image compression method according to claim 13, wherein the lossy compressed data is generated based on the upper limit generated information amount. 15. The lossy compression data is generated by performing lossless compression processing on the input image in block units to generate lossless compressed data, converting the input image into data on a frequency space and performing lossy compression processing. Generating, and selecting either the lossless compression data or the lossy compression data in the block unit based on the cumulative amount of generated information of the lossless compression data and the high frequency power value in the frequency space. Image compression method. 16. When generating the lossy compressed data, the image data is converted into a frequency space, the data in the frequency space is scanned, and the maximum value of the high frequency power is detected. Item 16. The image compression method according to Item 15. 17. An upper limit generation information amount allowable up to the currently processed block is calculated, the cumulative generation information amount is equal to or less than the upper limit generation information amount, and the maximum value of the high frequency power is a predetermined value. 17. The image compression method according to claim 16, wherein the lossless compression data is selected when the above is the case, and the lossy compression data is selected in other cases. 18. The image compression method according to claim 17, wherein the lossy compressed data is generated based on the upper limit generated information amount.