JPH05316356A - Picture signal decoder - Google Patents

Abstract

(57) [Summary] [Object] An object of the present invention is to provide an image restoration device capable of removing block distortion of image coded data using orthogonal transformation and improving visual deterioration. It is a thing. A coded image data is decoded to obtain orthogonal transform data, and a difference value between a DC component of each block and a DC component of the immediately preceding block horizontally adjacent to the block and a difference value between the block and A difference value with respect to the DC component of an adjacent block on the upper side is detected, and when the difference value is smaller than a predetermined value, the inverse of the block and the block immediately before or on the block adjacent to the block. The smoothing operation is performed only on the orthogonally transformed pixels. Further, the smoothing calculation is performed only on pixels in the vicinity of the boundary with the immediately preceding block or the adjacent block.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention restores an original image signal by decoding image-encoded data that has been encoded and compressed in order to shorten the transmission time of the image signal or reduce the storage capacity. Image signal restoration device.

[0002]

2. Description of the Related Art Conventionally, halftone image data such as a TV signal has an enormous amount of information, and image data for efficiently transmitting such image data or recording on a magnetic disk or an optical disk. There have been various attempts to compress the. As one of such image data compression methods,
It is well known to use orthogonal transformation of image data and entropy coding, and recently it is being standardized as a compression method of a color still image (JPEG method).

In this method, digital two-dimensional image data is divided into blocks each having an appropriate number of samples (for example, 8 × 8), and each block is subjected to a two-dimensional discrete cosine transform (DCT) which is a kind of orthogonal transform. Is performed, and a coefficient representing the energy component of the frequency composed of the DC component and the AC component is obtained,
Since energy is concentrated on a specific component by this conversion, a large energy component is finely quantized, while a small energy component is roughly quantized.

Further, Huffman coding, which is a kind of entropy coding, is performed on the quantized coefficient thus obtained. This method encodes while referring to a Huffman table that assigns codes with a code length according to the establishment of the appearance of quantized coefficients. A long code is assigned to reduce the code length for each block.

In order to restore the image coded data compressed in this way, a process reverse to the above is performed. That is, the original image signal can be restored by decoding the coded data using the Huffman table and inversely orthogonally transforming the decoded quantized coefficient.

[0006]

In the above method, it is possible to perform coding at a low bit rate by performing coarse quantization, but if such image coded data is restored, quantization error will occur. Due to the effect, an unnatural discontinuous state occurs at the block boundary called block distortion, and especially in a flat area where the density change of the image is small, the block distortion is conspicuous and has a visual adverse effect, which deteriorates the image quality. There is a problem called.

In order to improve such block distortion,
It has also been attempted to perform a smoothing operation on the entire restored image signal, but smoothing also blurs necessary contour information and high frequency components in the image, and it is not possible to effectively remove only block distortion. It was difficult.

The present invention is intended to solve the above problems,
It is an object of the present invention to provide an image signal restoration device capable of obtaining a restored image of good visual quality with no conspicuous block distortion even with low bit rate image coded data.

[0009]

To achieve the above object, a first invention is to decode coded image data to obtain orthogonal transform data, and means to inverse orthogonal transform the orthogonal transform data to obtain image data. A difference value between the DC component of each block of the orthogonal transform data and the DC component of the immediately preceding block horizontally adjacent to the block, and the DC component of the block immediately above the block vertically adjacent to the block. Means for obtaining a difference value of, and when the difference value is smaller than a predetermined value,
The block and the previous one or one adjacent to the block
And a means for performing a smoothing operation on the inversely-orthogonally-transformed pixels of the upper block.

Further, in the second invention, the means for performing the smoothing operation is only for pixels in the vicinity of the boundary with the immediately preceding block or the adjacent block.

[0011]

According to the present invention, in the above configuration, when the difference value of the DC component between the block and the block immediately before and one block adjacent to the block is obtained, and the difference value is smaller than the predetermined value. , The block and its adjacent block are blocks related to a flat area where the block distortion is easily noticeable and the density change of the image is small, and the block of the image signal obtained by the inverse orthogonal transform and its adjacent block By performing the smoothing operation only on the pixels, the smoothing process is performed only on the portion where the block distortion is conspicuous, and the visual deterioration can be improved without causing blurring in the entire image.

Further, according to the second aspect of the present invention, the smoothing operation is performed only on the pixels in the vicinity of the boundary line between the block and the adjacent block, so that only the block distortion is generated without blurring the necessary contour information in the block. Can be removed.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of an embodiment of an image signal decoding apparatus according to the present invention, showing only the main parts relating to the present invention and omitting the other parts.

In FIG. 1, a block of a specific size (hereinafter referred to as 8 × 8 in the present embodiment) is divided and subjected to orthogonal transformation, and after quantization, coded image coded data is decoded circuit 1 Is input to each block, is decoded for each block while referring to the decoding table 10, and is returned to the orthogonal transform data. The 8 × 8 = 64 pieces of orthogonal transform data for one block are stored in the block memory 2.

The orthogonal transform data stored in the block memory 2 is subjected to inverse orthogonal transform in the inverse orthogonal transform circuit 3, restored to the original image signal, and sequentially written in the frame memory 4 block by block. On the other hand, all DC components of the decoded orthogonal transformation data are sent to the DC component holding memory 5,
Retained. One DC component exists in one block and represents the average luminance level of the image in that block. It is located at the upper left end of the block in the orthogonal transform data.

FIG. 2 is a schematic diagram showing the block structure of orthogonal transform data, and the black circles in FIG. 2 represent DC components. The DC component of each of these blocks is two-dimensionally addressed and held in the DC component holding memory 5 like D (m, n) corresponding to the two-dimensional arrangement of the original image signal. Further, the DC difference circuit 6 detects the difference value between the DC components of the block immediately above and the block to the left of the adjacent block.

As shown in FIG. 2, the difference value Dy (m, n) between the DC component D (m, n) of a certain block and the DC component D (m-1, n) one level above and one to the left. DC component of D (m, n
The difference value Dx (m, n) from -1) is obtained by the following arithmetic expression.

Dy (m, n) = D (m, n) -D (m-1, n) Dx (m, n) = D (m, n) -D (m, n-1) Difference values Dy (m, n) and Dx obtained by
(M, n) is input to the smoothing calculation ON / OFF determination circuit 7, where a predetermined value α and Dy (m, n) and Dx (m,
n) are compared, and when α> Dy (m, n), D (m, n)
Blocks B (m, n) and D of the restored image data related to
The block B (m-1, n) relating to (m-1, n) is determined to be the smoothing operation ON, and the determination signal and the block address information are sent to the address generation circuit 8.

Similarly, when α> Dx (m, n), D (m,
n), the block B (m, n) of the restored image data and the block B (m, n-1) of D (m, n-1) are determined to be the smoothing operation ON, and the address generation circuit 8 receives the determination signal. And block address information. That is, α> D
When y (m, n) and α> Dx (m, n), B (m,
The blocks of (n), B (m-1, n) and B (m, n-1) are determined to be smoothing operation ON, and the determination signal and the block address information are sent to the address generation circuit 8.

The address generation circuit 8 is a smoothing calculation ON / OF.
In response to the ON / OFF determination signal from the F determination circuit 7 and the block address information, the address on the frame memory 4 of the pixel to be subjected to the smoothing calculation is generated, the target pixel is read from the frame memory 4, and the smoothing calculation circuit 9 is read. Send out.

FIGS. 3 and 4 are schematic diagrams of the pixel array of the image signal stored in the frame memory 4 and subjected to the inverse orthogonal transform, and show the regions to be smoothed. 8 × 8 blocks B (m, n), B (m-1, n), B (m, n)
-1) are DC components D (m,
n), D (m-1, n), and D (m, n-1).

FIG. 3 shows the first embodiment, and FIG. 4 shows the second embodiment. In FIG. 3A and FIG. 4A, α> Dy
When (m, n) and α ≦ Dx (m, n), FIG.
And FIG. 4B shows that α ≦ Dy (m, n) and α> Dx (m,
n), FIGS. 3C and 4C show α> Dy (m,
n) and α> Dx (m, n), the pixel to be smoothed is shown. The pixel in the thick frame is the target pixel.

In FIG. 3, smoothing calculation is performed on all pixels in the target block. In FIG. 4, the smoothing operation is performed only on the pixels adjacent to the boundary between the target blocks. In the embodiment of FIG. 4, an area having a width of 4 pixels located symmetrically with respect to the block boundary line is the target of the smoothing operation. Smoothing operation circuit 9
Various smoothing calculation methods can be considered, such as a simple arithmetic mean of target pixels or a weighted average of weighting specific pixels.

In the case of FIG. 3, it is possible to uniformly apply a 3 × 3 two-dimensional digital filter, which is often used as a smoothing filter, to the pixels in the target block.
The coefficients of the filter used are, for example, 1,2,1 2,4,2 2,1,2.

In the embodiment shown in FIG. 4, the smoothing calculation target pixel is simply averaged or weighted averaged among four one-dimensional pixels which are perpendicular to the block boundary, and the calculation result is calculated as two pixels adjacent to the block boundary, that is, in the figure. It is the data of the black circle part of 4.
The calculation result thus obtained is written again in the frame memory 4 and output as a restored image signal.

[0026]

As described above in detail, according to the present invention, the smoothing process is performed only on the flat area in the restored image, in which the density of the block change is notable and the density change is small.
Since the data of the part where the density change is large, which includes a lot of edge information and contour information, is stored without being smoothed, without causing unnecessary visual blurring in the entire image,
Block distortion can be removed.

Further, according to the second aspect of the present invention, the smoothing operation is performed only on the pixels in the vicinity of the boundary line between the block and its adjacent block, so that only the block distortion is generated without blurring the necessary contour information in the block. Can be removed, and the quality of the restored image can be improved.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an image restoration apparatus according to an embodiment of the present invention.

FIG. 2 is a schematic diagram showing a block structure of orthogonal transformation data of the image restoration device according to the present embodiment.

FIG. 3A is a schematic diagram of a pixel array showing a target region for smoothing calculation according to the first embodiment. FIG. 3B is a schematic diagram of a pixel array showing a target region for smoothing calculation. Is a schematic diagram of a pixel array showing the target area of the smoothing operation

FIG. 4A is a schematic diagram of a pixel array showing a target region of a smoothing operation according to the second embodiment. FIG. 4B is a schematic diagram of a pixel array showing a target region of the smoothing calculation. Is a schematic diagram of a pixel array showing the target area of the smoothing operation

[Explanation of symbols]

 1 Decoding Circuit 2 Block Memory 3 Inverse Orthogonal Transform Circuit 4 Frame Memory 5 DC Component Holding Memory 6 DC Difference Circuit 7 Smoothing Operation ON / OFF Judgment Circuit 8 Address Generation Circuit 9 Smoothing Operation Circuit 10 Decoding Table

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Yoshinori Amano 1006 Kadoma, Kadoma City, Osaka Prefecture Matsushita Electric Industrial Co., Ltd.

Claims (2)

[Claims] 1. An apparatus for restoring image data encoded by using orthogonal transformation of a block of a specific size, means for decoding the encoded image data to obtain orthogonal transformation data, and the orthogonal transformation data. And a means for obtaining image data by performing an inverse orthogonal transform on the block, and a difference value between the DC component of each block of the orthogonal transform data and the DC component of the immediately preceding block horizontally adjacent to the block and the vertical value of the block. Means for obtaining a difference value from the DC component of the block immediately above adjacent block, and, when the difference value is smaller than a predetermined value, the block and the block immediately before or one block above the block adjacent to the block. An image signal restoration device, comprising: means for performing a smoothing operation on the inversely transformed pixels. 2. The means for performing the smoothing operation is one before or one.
The image signal restoration device according to claim 1, wherein the image signal restoration device is applied only to pixels in the vicinity of the boundary with the adjacent adjacent block.