JPS63102476A - Picture compressing device - Google Patents

Abstract

PURPOSE:To reduce the occurrence of quantization errors when compressed picture data are produced, by dividing an original picture into plural blocks and subtracting the mean value of picture elements at the same spatial coordinates of each block from the corresponding picture element value in each block. CONSTITUTION:Under the control of an address control section 3, an original picture is divided into plural blocks and sent to an arithmetic section 4 where the calculation of pseudo ensemble zero average is executed by using a prescribed equation. The difference output obtained as the result of the calculation is fetched by a picture compressing means 7 as a pre-process output. Under the control of another address control section 12, a pre-process picture written in a memory is divided into prescribed blocks and sent to a 1st orthogonal transforming section 13. At the transforming section 13, each block is orthogonally transformed and sent to a rearranging section 17 where the blocks are rearranged by the same frequency components. Each of the rearranged blocks is orthogonally transformed at a 2nd orthogonal transforming section 20 and the transformed output is sent to a quantizing section 14 where the blocks are quantized.

Description

[Detailed description of the invention] [Purpose of the invention] (Industrial application field) The present invention is an image compression device, especially for encoding images.

The present invention relates to an image compression device that performs projection or conversion processing to store information while compressing apparent data.

(Prior Art) When transmitting and receiving a digital image through a communication line, when storing it in a storage device, or when reducing the load on the communication path or storage device, there is a method of expressing the original image with a smaller number of aRs, a so-called method. Image compression is required, and there are two types of image compression methods: reversible compression and irreversible compression.

In reversible compression, it is possible to completely restore the original image by using the run-length kneading method or the Huffman coding method compressed by predictive coding, etc., whereas KLT
, Irreversible compression such as cosine transformation can achieve a higher compression rate than reversible compression, but it is not possible to completely restore the original image, and the image quality deteriorates in proportion to the compression rate. It is necessary to limit the compression ratio to such an extent that no deterioration in surface quality is felt to the eyes.

Orthogonal transform, especially block orthogonal transform, is well known for lossy compression, and among these, block cosine transform is considered the most practical and promising.

FIG. 8 shows a conventional image compression device employing the block cosine transform method. As shown in the figure, this device includes a memory 11 that temporarily stores the captured original image;
An address control section 12 controls the address of this memory 11 and outputs a part of the image for each block;
The image forming apparatus includes an orthogonal transformation unit 13 that performs orthogonal transformation (for example, two-dimensional high-speed cosine transformation) on the original image within one block for each block, and a matrix conversion unit 14 that quantizes the output of this transformation. Quantization section 1
4 comprises a bit allocation table 15 and a quantizer 16 that quantizes the transform output of the orthogonal transform section 13 based on the table information.
The output is sent to an external device, such as a storage device, as compressed image data.

(Problems to be Solved by the Invention) However, in conventional devices, when an image compressed by the block orthogonal transformation method is restored, the joints of blocks are visible in the restored image (referred to as block effect).
, the image quality will deteriorate. This phenomenon is caused by a q-conversion error in low-frequency components of the image, and becomes more pronounced as the compression rate increases.

The present invention has been made in view of the above circumstances, and an object thereof is to provide an image compression device that reduces quantization errors in generating compressed image data using a block orthogonal transform method.

[Structure of the Invention] (Means for Solving the Problems) The present invention provides a method for converting an original image into a plurality of blocks before an image compression means that performs image compression by quantizing image data after block orthogonal transformation. A preprocessing means is provided for preprocessing the original image data by dividing the image data and subtracting the average value of pixels at the same spatial coordinates of each block from the corresponding pixel value in each block.

(Function) One of the causes of increasing quantization errors is that, as shown in Figure 5, for example, the amplitude distribution of the same frequency component after orthogonal transformation deviates from the center set in the generator, causing quantization This is because more components protrude from the band. The inventors of the present application took note of this and provided the above-mentioned preprocessing means before the image compression means, so that the preprocessing result of the original image data is sent to the image compression means.

That is, the preprocessing means divides the captured original image into a plurality of blocks (these are referred to as sub-blocks), calculates the average of pixels at the same spatial coordinates of each block (pseudo unsampled average), and calculates this average value. is subtracted from the corresponding pixel value in each block (this is called pseudo unsampled zero average), and the output of this calculation is sent to the image compression means. Here, the unsample average value M eam is the seventh
As shown in the figure, when the pixel value at the spatial coordinate r in the i-th (i is a positive integer) sub-block is f(r, ωi), it is expressed as follows. However, \ means the total number of blocks.

In addition, the pseudo unsampling zero average value f' (r.

ωi) is f' (r, ω1)=f (r, ωi)−Mea
It is expressed as n(r) (2), and the following relationship holds true between the pseudo unsampled zero average values. Then, the calculation result (preprocessing output) according to the above equation (2) is sent to the image compression means and subjected to image compression.

By preprocessing the original image data in this way, the distribution of amplitudes of the same frequency components subjected to block orthogonal transformation becomes as shown in FIG. 6, and does not deviate from the center of the quantization band. This reduces quantization errors and
Image compression stability and S/N ratio can be improved.

(Example) Hereinafter, the present invention will be specifically explained with reference to Examples.

FIG. 1 is a block diagram of an image compression apparatus which is an embodiment of the present invention. As shown in the figure, the device of this embodiment divides the original image into a plurality of blocks, and subtracts the average value of pixels at the same spatial coordinates of each block from the value of the corresponding pixel in each block. It comprises a preprocessing means 1 that performs preprocessing, and an image compression means 7 that is arranged after the preprocessing means 1 and performs image compression by quantizing the preprocessing output after block orthogonal transformation.

The preprocessing means 1 includes a memory 2. Address control section 3° calculation section 4. Memory 5. It has a compression encoding section 6.

The memory 2 stores original image data taken in from outside, and the address control section 3 divides the original image into a plurality of blocks by controlling the addresses of this memory. The calculation unit 4 obtains a pseudo unsampled average value and a pseudo unsampled zero average value by executing the calculations of the preceding equations (1) and (2). It is sent to the conversion section 6,
Further, the pseudo unsampled zero average value is sent to the image compression means 7. The compression encoding unit 6 compresses and encodes the captured pseudo unsampled average value, and its output is sent to the outside as head information of compressed image data.

The image compression means 7 also includes a memory 11. Address control unit 12. First orthogonal transform unit 13. Relocation unit 17, second orthogonal transformation unit 20. It has a mothering part 14. The memory 11 temporarily stores the preprocessed output, the address controller 12 outputs the preprocessed image block by block by controlling the address of the memory 1, and the first orthogonal transform unit 13 stores the preprocessed output for each block. The preprocessed image output from block 11 is orthogonally transformed (for example, two-dimensional fast cosine transform) for each block, and both have the same functions as shown in FIG.

The rearrangement unit 17 rearranges the transform output of the first orthogonal transform unit 13 into blocks of the same frequency component, and includes a memory 19 and an address controller 18 that controls the address of this memory 19. It consists of Second orthogonal transform unit 2
0 performs orthogonal transformation for each block rearranged by this rearrangement unit 17. Note that this second orthogonal transformation section 20 performs two-dimensional (2-D) or one-dimensional (1-D
) is performed.

The mother-child conversion unit 14 quantizes the transform output of the second orthogonal transform unit 20, and includes a bit allocation table 15 and a quantizer 16 as shown in FIG.
It consists of

Next, the operation of the above configuration will be explained according to the flowchart of FIG. 2.

First, an original image is divided into a plurality of blocks under the control of the address control section 3 (step S1) and sent to the calculation section 4. Then, in the calculation section 4, the calculation processing of equations (1) and (2) (calculation of pseudo unsampled zero average) is performed (step S2). The average value obtained by executing the calculation of equation (1) above is sent to the compression encoding unit 6 via the memory 5, where it is compressed and encoded, and then output as head information (step S7).

On the other hand, the difference output obtained by executing the calculation of equation (2) in the calculation section 4 is taken into the image compression means 7 as a preprocessing output. Then, the preprocessed image written in the memory 11 under the control of the address control unit 12 (FIG. 3(a)
) is again controlled by the address control unit 12 as shown in FIG.
As shown in b), the signal is divided into predetermined blocks and sent to the first orthogonal transform unit 13. Then, the first orthogonal transform unit 13 performs orthogonal transform (two-dimensional fast cosine transform) for each block as shown in FIG. 3(C) (step S
3), this converted output is sent to the relocation unit 17. This rearrangement unit 17 receives i! of the address controller 18. By control, it is rearranged into blocks for each same frequency component (
Step 83). The orthogonal transform output in FIG. 3(C) is roughly expressed in detail as shown in FIG. 3(d), which is rearranged into blocks for each same frequency component as shown in FIG. 3(e).

In addition, the difference in circles in the figure means the difference in frequency.

Then, for each rearranged block, the second orthogonal transform unit 2
0, orthogonal transformation (two-dimensional or one-dimensional high-speed cosine transformation) is performed (step S5), and the output of this transformation is sent to the quantization unit 14, where it is mother-sonified (step S3).
6). FIG. 3 (f> shows the transform output from the second orthogonal transform section 20.

In this way, in the apparatus of this embodiment, the preprocessing means 1 is provided before the image compression means 7 to compress the preprocessing output, so that the amplitude distribution of the same frequency component that has been subjected to block orthogonal transformation is , the quantization band will not deviate from the center, quantum errors can be reduced, and image compression stability and S/N ratio can be improved.

In addition, the original image is orthogonally transformed block by block, the transform output is rearranged into blocks of the same frequency component, orthogonal transform is performed for each rearranged block, and this transform output is converted into a matrix. This method does not transform the original image block by block and quantize it immediately after orthogonally transforming the original image, as in the past, so it is possible to concentrate the power of the image more than before, and by doing this, it is possible to concentrate the power of the image more than before. It is possible to further reduce errors caused by Therefore, when compressed image data generated by the device of this embodiment is restored, "block seams" caused by quantization errors are reduced, and S
By improving the /N ratio, images with excellent diagnostic performance can be obtained. In addition, at the same compression rate, quantization can be performed using fewer bits and with smaller errors compared to conventional equipment.
A large number of bits can be allocated to high frequency components,
This also has the advantage of increasing the spatial resolution and density resolution of the restored image. Furthermore, since a restored image of the same quality as before can be obtained with a smaller number of bits,
It is also possible to increase the compression ratio compared to the conventional method, thereby shortening the transfer time of compressed image data and saving storage capacity.

Although one embodiment of the present invention has been described above, it goes without saying that the present invention is not limited to the above-mentioned embodiment and can be implemented in various modifications.

For example, in the above embodiment, the second orthogonal transformation unit 20 performs orthogonal transformation on all blocks rearranged by the rearrangement unit 17, but the rearranged blocks do not require orthogonal transformation. Since blocks also exist, it is preferable to send such blocks to the quantization unit 14 without passing through the second orthogonal transformation unit 20. This will be more advantageous than the above embodiments in terms of image quality and processing time reduction. Image compression means 7A in this case
An example of the configuration is shown in FIG. In the figure, relocation unit 1
A block selector 21 is provided between the block selector 7 and the second orthogonal transform unit 20.
The block selector 21 divides the blocks into blocks that require orthogonal transformation by the second orthogonal transformation unit 20 and blocks that do not. Whether or not orthogonal transformation is required can be clearly determined from the address of the rearranged block, so block selection can be easily achieved by controlling the address. Furthermore, the bit allocation table of the quantization unit is also switched as necessary.

[Effects of the Invention] As described in detail above, according to the present invention, it is possible to provide an image compression device that aims to reduce quantization errors in generating compressed image data using a block orthogonal transform method.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a flowchart for explaining the operation of the device of this embodiment,
FIGS. 3(a) to 7(f) are explanatory diagrams of various processes in the apparatus of this embodiment, FIG. 4 is a block diagram showing one embodiment of the present invention, and FIGS. 5 to 7 are diagrams of the present invention. Principle explanatory diagram, No. 8
The figure is a block diagram of a conventional example. 1... Preprocessing means, 7, 7A... Image compression means, 1
3... First orthogonal transformation unit, 14... Quantization unit, 17
... Relocation unit, 20... Second orthogonal transformation unit. Figure 6

Claims (2)

[Claims] (1) In an image compression device comprising an image compression means that performs image compression by quantizing image data after block orthogonal transformation, an original image is divided into a plurality of blocks, and each block has the same spatial coordinates. An image compression device comprising: preprocessing means for preprocessing original image data by subtracting the average value of pixels from the value of corresponding pixels in each block, provided at a stage upstream of the image processing means. (2) The image compression means includes a first orthogonal transformation section that orthogonally transforms the original image block by block, a relocation section that rearranges the orthogonal transformation output into blocks of the same frequency component, 2. The image compression device according to claim 1, comprising: a second orthogonal transform unit that performs orthogonal transform on a block-by-block basis; and a quantizer that quantizes the orthogonal transform output.