JP3036071B2 - Image compression device - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to an image processing technique including image compression, and in particular,
The present invention relates to an image compression device suitable for a high-quality image compression device in a narrow transmission band, such as a high-quality image compression device using DCT (Discrete Cosine Transform).

[Prior Art] In a conventional image compression apparatus of this type, generally, as shown in FIG. 3, input image data of one frame to be processed is not overlapped with 8 dots × 8 lines as one block. Each block is subjected to DCT processing and image compression. In this case, processing is sequentially performed individually for each block of 8 dots × 8 lines, for example, from left to right on the image plane, from top to bottom.

[Problems to be Solved by the Invention] In the conventional image compression apparatus described above, for example, 8 dots ×
DCT processing is performed individually for each block of 8 lines, and processing of each block has no correlation with adjacent blocks. For this reason, the image output as the processing result is likely to be an image in which a tiled mosaic pattern not included in the original image is generated, that is, a mosaic image. Therefore, the conventional image compression apparatus has a disadvantage that the image quality is significantly deteriorated by the compression processing. In particular, such image compression processing using DCT processing or the like is indispensable when transmitting high-quality image data through a transmission system with a narrow transmission band. However, the fact that the image quality is degraded by image compression for transmission has been a serious problem.

The present invention has been made in view of such a problem, and an object of the present invention is to provide an image compression apparatus in which image quality is hardly degraded due to image data compression processing.

[Means for Solving the Problems] An image compression apparatus according to the present invention includes a first memory for storing at least one frame of input image data, and compressing the input image data for each block of a predetermined number of dots × the number of lines. Compression processing means for processing; a second memory for storing the compressed image data; and a first memory for storing the image data for each block having the predetermined number of dots × the number of lines and overlapping each other at a boundary portion. Processing control to supply the image data of each block, which has been compressed by the compression processing means, to the second memory in correspondence with the address in the first memory. Means for mutually overlapping image data corresponding to the overlapping portion of the image data for each block stored in the second memory. Anda calculation means for performing arithmetic processing, the processing control unit,
When duplicate data at the same address is transferred to the second memory, the data already written at the corresponding address of the second memory is read and sent to the arithmetic means. The calculated data and the newly transferred data are calculated so that they have a correlation.

Specifically, for example, DCT processing is performed for each block of 16 dots × 16 lines, and the block to be processed is overlapped with an adjacent block by an addresser, and the processed data of the overlapping portion is averaged. Perform arithmetic processing such as taking a value. For this reason, the correlation between the block serving as the processing unit and the adjacent block is high, and the processed output image is unlikely to be a mosaic image including a tile pattern.

[Operation] In the image compression apparatus according to the present invention, the blocks serving as processing units are overlapped with each other at the boundary between the blocks, and the result of the compression processing is mutually calculated for the overlapping portion due to the overlap. , And a tiled mosaic pattern is less likely to occur in the processed output image.

Embodiment An embodiment of the present invention will be described below with reference to the accompanying drawings.

FIG. 1 shows a configuration of an image compression apparatus according to one embodiment of the present invention.

The image compression apparatus shown in FIG.
It includes a CT processing unit 2, an addresser 4, and a calculation unit 5.

In the first frame memory 1, only one frame of input image data to be processed is stored. DCT processing unit 2
Performs compression processing by DCT (discrete cosine transform) on image data for each block of 16 dots × 16 lines read from one frame memory 1 as specified by the addresser 4. The image data compressed by the DCT processing unit 2 is sequentially stored in the second frame memory 3 according to the designation of the addresser 4. Addresser 4 is 16 dots x 16
When the image data for each block of the line is read from the first frame memory 1, each block to be read is, for example, an image in which an overlap portion of 4 dot width and 4 line width is generated as shown in FIG. Specify sequentially from left to right on the plane and from top to bottom. The addresser 4 is 16 dots × 16 pixels processed by the DCT processing unit 2.
The image data for each block of the line is written to an address of the second frame memory 3 corresponding to the read address in the first frame memory 1. At this time, the DCT processing unit 2
The data of the overlapped portion due to the overlap of the image data of each block of 16 dots × 16 lines processed in
The arithmetic unit 5 takes, for example, an average value when writing the data into the frame memory 3. Therefore, the DC is stored in the second frame memory 3.
Image data that has been processed by the T processing unit 2 and that has been further averaged for the overlapping portion of each block and given a correlation with an adjacent block is stored. The arithmetic processing of the average value in the arithmetic unit 5 is performed as follows. That is, the previously processed data is once written in the second frame memory 3, and when duplicate data is transferred to the same address, the data is already written to the corresponding address in the second frame memory 3. The read data is read, the arithmetic unit 5 calculates an average value of the read data and the newly transferred data, and writes the result to a corresponding address of the second frame memory 3.

Conventionally, generally, as shown in FIG. 3, 8 dots × 8 lines are regarded as one block without correlation with adjacent blocks.
Since the DCT processing was performed, the mosaic pattern was likely to appear in the output image after the processing, but in the present embodiment, as shown in FIG. DCT processing is performed. Then, the image data of the overlapping portion takes the average value after the DCT processing. As a result, the image data of each block has a high correlation with the adjacent blocks, and the output image obtained as a result of the processing has a smooth block boundary, and the deterioration of the image quality is effectively suppressed.

The present invention is not limited to the above-described embodiment, and can be implemented with various modifications. For example, as image compression processing, DC
Compression processing other than T processing can also be used. Further, one or more frames of image data may be stored in the frame memory 1 or 3. Further, a block other than 16 dots × 16 lines may be used as a unit of the image data to be compressed, and the width of the overlap portion of the adjacent block may be a value other than 4 dots and 4 lines. Also,
Since it is only necessary to make the overlapping portion have a correlation as the arithmetic processing of the overlapping portion, weighted addition or geometric mean may be used in addition to the arithmetic mean.

[Effects of the Invention] As described above, according to the present invention, since adjacent blocks are processed in an overlapping manner, adjacent blocks have a correlation and block boundaries are smoothed, thereby suppressing deterioration in image quality due to image compression. An image compression apparatus capable of performing the above-described operations can be provided.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an image compression apparatus according to an embodiment of the present invention, FIG. 2 is a view for explaining an image compression processing method in the embodiment, and FIG. 3 is a conventional general image compression processing. It is a figure for explaining a method. 1,3; Frame memory, 2; DCT processing unit, 4; Addresser, 5; Operation unit

Claims (1)

(57) [Claims] 1. A first memory for storing at least one frame of input image data, compression processing means for compressing the input image data for each block of a predetermined number of dots × the number of lines, and image data compressed And a second memory for storing image data for each block of the predetermined number of dots × the number of lines and overlapping each other at a boundary portion from the first memory to the compression processing means. Processing control means for controlling the image data of each block, which has been subjected to the compression processing, to be stored in the second memory in correspondence with the address in the first memory, and the block stored in the second memory Computing means for performing a predetermined computing process on the overlapping image data corresponding to the overlapped portion among the image data for each of the image data. When duplicate data at the same address is transferred to the second memory, the logical control unit reads the data already written at the corresponding address of the second memory, sends the data to the arithmetic unit, and Means for calculating the read data and the newly transferred data such that the two have a correlation with each other.