JPH06165149A - Image coding control method - Google Patents

Abstract

(57) [Summary] [Object] For a block in which mosquito noise is likely to occur, a coding parameter is controlled so that the code amount is larger than that of other blocks, and a decoded image with little deterioration in image quality is obtained. [Structure] When encoding, a mosquito noise occurrence determination unit 6 determines each block (N × N pixels) of an input image.
For, it is determined whether mosquito noise is likely to occur. The quantization step width modulator 7 controls the block for which it is determined that mosquito noise is likely to occur, by reducing the quantization step width and performing quantization.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image coding method in which a moving image is orthogonally transformed and coded in block units of m × n pixels, and occurs in a decoded image when coding at a high compression rate. The present invention relates to an image coding control method that suppresses mosquito noise.

[0002]

2. Description of the Related Art In recent years, as a compression coding method for moving images,
The MPEG method has been proposed. FIG. 6 is a block diagram of a conventional image encoder. The input image is DCT-transformed in a block unit of N × N pixels in the orthogonal transformer 1, the transform coefficient is quantized in the quantizer 2, and variable-length coded (Huffman coded) in the VLC unit 3, which is not shown. Stored in the buffer. The encoded data stored in the buffer is read at a constant rate and sent to a digital storage medium or the like.

As described above, in the image coding method for orthogonally transforming and coding a moving image in block units, coding parameters (for example, such that the average generated code amount in a certain range becomes a predetermined bit rate). Quantization step size)
Need to control.

That is, in FIG. 6, conventionally, the cumulative amount of the difference between the target code amount and the actual generated code amount (buffer occupancy amount) is obtained by the counter 4, and the quantization step width is determined according to the buffer occupancy amount. The deciding unit 5 controls the quantization step width in units of a plurality of blocks.

FIG. 7 is a diagram for explaining the quantization step width control based on the buffer occupancy amount. For example, when the buffer occupancy amount is the maximum value of the buffer, the quantization step width is the maximum value Qmax and the buffer occupancy amount is When 0, the quantization step width is determined in proportion to the buffer occupancy so that the quantization step width becomes the minimum value Qmin. Then, for example, at the position of * in the figure, since the buffer occupancy amount is 3/5 of the buffer maximum value Qmax, the quantization step width used in the encoding of the next block is 3/5 ×
Let Qmax. In this way, the average generated code amount is controlled so as to be close to the predetermined bit rate.

[0006]

In the above-mentioned conventional coding method, the coding parameter (quantization step width) is controlled by considering only the control of the generated code amount regardless of the occurrence of coding noise. Therefore, if the bit rate is controlled to a low bit rate (that is, a high compression rate), coding noise, particularly mosquito noise, is generated in some blocks, which causes a problem that the quality of a decoded image is significantly deteriorated.

An object of the present invention is to control a coding parameter for a block in which mosquito noise is likely to occur so that the code amount is larger than that of other blocks, and to obtain a decoded image with little deterioration in image quality. Another object of the present invention is to provide an image coding control method.

[0008]

In order to achieve the above object, according to the invention of claim 1, when a moving picture is coded in block units, a quantization step width is changed in a unit of a plurality of blocks and a code amount is changed. In the image coding control method for controlling, the quantization step width is controlled to be small according to the easiness of occurrence of coding noise in the block.

According to the second aspect of the present invention, the easiness of occurrence of the coding noise is determined by setting a window of 3 × 3 pixels for each pixel in the block and adjoining the pixels around the central pixel. The absolute value of the gradation level difference from the pixel is calculated, and the number of pixels whose ratio between the average value and the range width of the gradation level signal in the block is less than or equal to a predetermined threshold is calculated. Is determined by whether or not there are a predetermined number or more in the block.

[0010]

In order to determine a block in which mosquito noise is likely to occur, a window of 3 × 3 pixels is applied to each pixel in the block, and the surrounding 8 pixels surrounding the central pixel are
The absolute value of the gradation level difference between each adjacent pixel is averaged, and the ratio between this value and the range width of the gradation level signal in the block is obtained. Then, the number of pixels whose ratio is equal to or less than a predetermined threshold value is counted in the block. A block whose number exceeds a certain threshold is determined to be a block in which mosquito noise is likely to occur. That is, a block including a large edge in a block and having many flat portions is determined as a block in which mosquito noise is likely to occur.

For a block determined to be a block in which mosquito noise is likely to occur in this way, the quantization step width in the conventional coding control is multiplied by K (0 <K <
By applying the item 1), the deterioration of the image quality in these blocks can be suppressed.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be specifically described below with reference to the drawings. FIG. 1 is a block diagram of the image encoder of the present invention. The encoder of the present invention has a conventional configuration,
The configuration is such that a mosquito noise occurrence determination unit 6 and a quantization step width modulator 7 whose quantization step width is modulated according to the determination result of the determination unit are added.

By the way, the mosquito noise is a low-order portion of a flat portion around an edge portion of the image, which is compressed when an edge portion of the image such as a character or a contour whose gradation changes drastically affects a processing block of the orthogonal transformation. This is a confusing noise that occurs due to the quantization error that occurs in the AC component of. FIG. 5A is a diagram showing the edge portion of the original image, and FIG. 5B is a diagram showing the edge portion of the decoded image having mosquito noise.

According to the present invention, a mosquito noise easiness determination unit is provided in order to determine a block in which such mosquito noise easily occurs. FIG. 2 is a flowchart of the determination process of the mosquito noise occurrence easiness determination unit.

In FIG. 2, a square block of N × N pixels is input to the mosquito noise susceptibility determining section (step 101). The gradation level of each pixel in the block is represented by X (j, i) (j, i = 0, ..., N-1).

First, the maximum gradation level difference Δ of the target block
Xmax (= Max-Min) is calculated (step 10).
2).

ΔXmax = MaxX (j, i) -Min
X (j, i) (however, ΔXmax is calculated for j and i from 0 to N−1).

Then, each pixel in the block is cut out by applying a window of 3 × 3 pixels centered on the pixel (step 103). FIG. 3B shows a window of 3 × 3 pixels,
(A) is 3 × 3 cut out from the N × N pixel block
Indicates a pixel. The cut-out array of the gradation levels of 3 × 3 pixels is set to W (k, m) (k, m = 0,1,2). Figure 4
Indicates an array of 3 × 3 pixel gradation levels.

Then, in the cut out 3 × 3 pixels,
Flatness F of gradation level around the central pixel (target pixel)
(J, i) is evaluated by averaging the absolute values of the gradation level differences between the adjacent 8 pixels around the central pixel (step 104).

That is, F (j, i) = (| W (0,0) -W (0,1) | + | W (0,1) -W (0,2) | + | W (0, 2) -W (1,2) | + | W (1,2) -W (2,2) | + | W (2,2) -W (2,1) | + | W (2,1) -W (2,0) | + | W (2,0) -W (1,0) | + | W (1,0) -W (0,0) |) / 8 is calculated.

It is determined whether or not the ratio of the flatness evaluation value F (j, i) to the previously determined maximum gray level difference ΔXmax in the block is less than or equal to a predetermined threshold value k1 (step 105), it is determined that pixels having a threshold value k1 or less may have mosquito noise.

The number of pixels (Nf
Are counted in the block (steps 106, 1)
07). A block in which Nf exceeds a predetermined threshold value k2 is determined as a block in which mosquito noise is likely to occur (steps 108 and 109). That is, a block including a large edge in a block and having many flat portions is determined as a block in which mosquito noise is likely to occur.

Based on the result of the determination, the quantization step width modulator 7 has a quantization step width Q × K obtained by multiplying the quantization step width Q by K times (0 <K <1) for such a block. To the quantizer 2.

As described above, according to the present invention, it is determined in advance whether noise is likely to occur during the encoding process, and the quantization step width is reduced for the block that is determined to be susceptible to noise. It is controlled to suppress the generation of noise. As a result, a larger number of bits is allocated to the block determined to be likely to generate noise than the conventional method, and conversely, a block determined to be less likely to generate noise is
A smaller number of bits will be allocated than in the conventional method. Therefore, while the image quality of the decoded image of the block determined to be likely to generate noise is improved, the image quality of the block determined to be less likely to generate noise is reduced. Since it is smaller than the improvement of, the image quality will be greatly improved as a whole.

[0025]

As described above, according to the first and second aspects of the present invention, the quantization step width is reduced for the block where coding noise is likely to occur. It is possible to suppress the occurrence of mosquito noise in a decoded image at a low bit rate, as compared with the coding control method.

[Brief description of drawings]

FIG. 1 is a block configuration diagram of an image encoder of the present invention.

FIG. 2 is a flowchart of a determination process of a mosquito noise occurrence easiness determination unit.

3A shows 3 × 3 pixels cut out from an N × N pixel block, and FIG. 3B shows a window of 3 × 3 pixels.

FIG. 4 shows an array of 3 × 3 pixel gradation levels.

FIG. 5A shows an edge portion of an original image, and FIG.
FIG. 9 is a diagram showing an edge portion of a decoded image in which mosquito noise has occurred.

FIG. 6 is a block configuration diagram of a conventional image encoder.

FIG. 7 is a diagram illustrating a quantization step width control based on a buffer occupation amount.

[Description of Codes] 1 Orthogonal transformer 2 Quantizer 3 VLC device 4 Cumulative code amount counter 5 Quantization step width determination unit 6 Mosquito noise occurrence determination unit 7 Quantization step width modulator

Claims (2)

[Claims] 1. An image coding control method for controlling a code amount by changing a quantization step width in a unit of a plurality of blocks when a moving image is coded in a unit of a block. The image coding control method is characterized in that the quantization step width is controlled to be small according to the size. 2. The determination of the easiness of occurrence of the coding noise is performed by setting a window of 3 × 3 pixels for each pixel in the block, and gradation levels of adjacent pixels of pixels around the central pixel. Obtaining the average value of the absolute values of the differences, with the average value,
The number of pixels whose ratio to the range width of the gradation level signal in the block is less than or equal to a predetermined threshold is calculated, and the number of pixels is determined by whether or not the number of pixels is greater than or equal to a predetermined number. The image coding control method according to Item 1.