JP2758789B2 - Image coding device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image coding apparatus for coding digital image data.

[0002]

2. Description of the Related Art In recent years, an image coding apparatus has been proposed which uses a coding method in which a digital image is divided into blocks within a certain range such as a frame or a field and the amount of information is adaptively allocated to each block. .

FIG. 3 is a block diagram of a conventional image coding apparatus.
Shown in First, digital image data of one screen unit, that is, one frame unit or one field unit is input to an image encoding device (not shown).

In the blocking section 100, as shown in FIG. 2, image data is divided into blocks 10 each having a total of 64 pixels of 8 pixels by 8 pixels and output as image data of each block 10. That is, one frame or one
The screen 11, which is constituted by fields and has a total of 802,560 pixels of 240 pixels vertically × 704 pixels horizontally,
X 88 blocks divided into a total of 2640 blocks 10

The fixed-length block forming unit 101 uniformly collects a predetermined number of image data for each block 10 received from the blocking unit 100, allocates a fixed number of bits, and outputs the fixed-length block image data. I do. However, in this case, each block 10 corresponds to the screen 1
Select and collect them so that they are not adjacent to each other in 1.

In the orthogonal transform unit 102, orthogonal transform is performed by DCT, Hadamard transform or the like for each block 10 belonging to each fixed-length block received from the fixed-length block forming unit 101, and an orthogonal transform coefficient of 64-pixel image data is obtained. Output. The same output is performed for another block included in the fixed-length block being processed, and the same process is repeated for the other fixed-length blocks to perform orthogonal transformation of all pixel data.

The classifying section 103 calculates the variance or dynamic range of the image data of the fixed-length blocks, and combines the image data of the fixed-length blocks having substantially the same magnitude in the calculation result into one class, and all the fixed-length blocks. Image data into several classes.

[0008] The bit allocation unit 104 allocates the number of bits required for each class divided by the classifying unit 103 within a range not exceeding the number of bits allocated to image data allocated to each fixed-length block. For example, a small number of bits is assigned to class data of a class with a small variance, and a large number of bits is assigned to class data of a large class.

[0009] The quantization unit 105 includes a bit allocation unit 10
4 quantizes the orthogonal transform coefficient calculated from the 64 pixel data of the block belonging to one fixed-length block with the number of bits assigned to the class to which the image data of each fixed-length block belongs.

[0010] The encoding unit 106 receives an input of orthogonal transform coefficients of 64 pixel data of a block belonging to the fixed-length block quantized by the quantization unit 105 and encodes it.

[0011]

However, when an image as shown in FIG. 4 is inputted as described above, since the edge portion 15 includes many boundary lines 17 and irregularities 18, the variance and the dynamic range are reduced. Image data of a large background 13 and flat portion 1 with few boundaries 17 'and unevenness 18'
Because of 6, person 1 with small variance and dynamic range
In a state in which the image data of the four blocks are mixed, the fixed-length block forming unit 101 forms one fixed-length block from a plurality of non-adjacent blocks. 13 image data,
Image data of the person 14 that is the flat portion 16 is mixed.

Therefore, the edge portion 1 in one fixed-length block
For the image data of the background 13 which is 5, the variance is large, so that a large number of bits are allocated. For the image data of the person 14 which is the flat portion 16, the number of bits determined by one fixed-length block is In order not to exceed
Not enough bits are allocated, resulting in a flat
The distortion of the person 14 of No. 6 was large.

An object of the present invention is to provide an image coding apparatus which solves such a problem and allocates appropriate bits to reduce image distortion.

[0014]

In order to achieve the above object, an image coding apparatus according to the present invention comprises a predetermined number of adjacent pixels collectively formed as one block, which is composed of image data of a plurality of pixels. The image data of one screen is divided into image data of a plurality of blocks, and
In an image coding apparatus that forms a fixed-length block and orthogonally transforms all pixel data in each block belonging to one fixed-length block to obtain an orthogonal transform coefficient, and quantizes and encodes this coefficient, An area dividing unit that divides a screen by the area by grouping matching blocks together, and an area in which the result can be regarded as substantially the same based on a statistical result of an image forming characteristic of image data of a block belonging to each area. Is included in this group as a group, and even in the area not included, the area that can be regarded as having almost the same result is repeatedly included as a group, thereby dividing the area into several groups in units. Grouping means, and each group formed by the grouping means has And wherein the configuring the assignment 1 fixed length block the partial number of bits.

[0015]

According to the present invention, based on the statistical results of the characteristics related to the image formation of the image data of the blocks belonging to the respective regions, regions that can be regarded as having substantially the same result are included in one group, so that one screen can be divided into several screens. Divided into groups. Accordingly, it is possible to limit the processing of forming one fixed-length block and performing the processing on the image data within each group having substantially uniform characteristics related to image formation.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below in detail with reference to the drawings. FIG. 1 is an example of a block diagram showing an embodiment of the present invention. FIG. 2 shows an example of a method for dividing blocks and areas in one screen, that is, one frame or one field, according to the present invention.

In FIG. 2, reference numeral 10 denotes a block composed of a total of 64 vertical 8 pixels × 8 horizontal pixels.
This area is composed of a total of 220 blocks × 22 horizontal blocks. That is, one area 12 is composed of a total of 14080 pixels in a vertical direction and 80 pixels in a horizontal direction. The screen 11 is divided into a total of twelve areas 3 × 4 × 12.

In FIG. 1, first, digital image data is input to an image encoding device (not shown) for each image, that is, for each frame or each field. The blocking unit 1 converts the input digital image data into eight vertical pixels ×
The image data is divided into blocks 10 each having a total of 64 horizontal 8 pixels, and each image data is output.

The area dividing section 2 divides one screen, that is, one field or one frame image, as shown in FIG.
The block is divided into 12 blocks (3 blocks × 4 blocks) by an area 12 composed of 10 blocks vertically × 22 blocks horizontally, ie, a total of 220 blocks 10, that is, a total of 80 pixels vertically × 176 pixels horizontally, and a total of 14080 pixels.

The grouping section 3 first has 8 vertical pixels × 8 horizontal pixels.
The dynamic range of the image data of each block 10 including the pixel total 64 is calculated, and the sum of the dynamic ranges of the image data of all the blocks 10 belonging to the area 12 is obtained. Between each region 12, each region 1
The sum of the dynamic ranges of the image data of all the blocks 10 belonging to 2 is compared, and if the difference is large, they are grouped. If there is no difference, the whole is regarded as one group.

For example, since the edge portion 15 shown in FIG. 4 includes a large number of boundaries 17 and irregularities 18, the sum of the dynamic ranges becomes large. Therefore, the image data corresponding to the edge portion 15 shown in FIG. 12b, 12c, 1
2e, 12f, 12g, 12i, 12j, and 12k are collectively referred to as a group G1. Further, the flat portion 16 shown in FIG.
Since the sum of the dynamic ranges is small because there are few boundary lines 17 ′ and unevenness 18 ′, the regions 12 d, 12 h, shown in FIG.
12l are collectively referred to as group G2. Thus, the grouping section 3 divides the twelve regions 12 into several groups.

Thereafter, the fixed-length blocking unit 4, the orthogonal transformation unit 5, the classification unit 6, the bit allocation unit 7, and the quantization unit 8
The processing in the encoding unit 9 is conventionally performed collectively for each screen, but is performed separately for each group.

The fixed-length block forming unit 4 uniformly collects a predetermined number of image data of the block 10, allocates a fixed number of bits, and outputs the data as one fixed-length block of image data. However, in this case, the blocks 10 are selected so as not to be adjacent to each other in the group being processed.

In the orthogonal transform unit 5, each pixel data is orthogonally transformed by DCT, Hadamard transform or the like for each block 10 belonging to each fixed-length block received from the fixed-length block forming unit 4 to obtain 64 pixel data. Output orthogonal transform coefficients. The same output is performed for another block included in the fixed-length block being processed, and the same process is repeated for the other fixed-length blocks to perform orthogonal transformation of all pixel data.

The classifying unit 6 calculates the variance or dynamic range of the image data for each fixed-length block, and combines the image data of the fixed-length blocks having substantially the same magnitude in the calculation result into one class, and forms all fixed-length blocks. Divide the image data of the block into several classes.

The bit allocating unit 7 allocates the number of bits required for each class divided by the classifying unit 6 within a range not exceeding the number of bits allocated to each image data allocated to each fixed-length block. . For example, a small number of bits is assigned to class data of a class with a small variance, and a large number of bits is assigned to class data of a large class.

The quantization unit 8 calculates the number of bits assigned to the class to which the image data of the fixed-length block belongs by the bit allocation unit 7 and from the 64 pixel data of the block belonging to one fixed-length block. The quantized orthogonal transform coefficients are quantized. The encoding unit 9 receives an input of the orthogonal transform coefficient of the image data belonging to the fixed-length block quantized by the quantization unit 8 and encodes the same.

[0028]

As described above, according to the image encoding apparatus of the present invention, the results can be considered to be substantially the same based on the statistical results of the image formation characteristics of the image data of the blocks belonging to each area. By including the area in one group, one screen is divided into several groups. Accordingly, it is possible to limit the processing of generating one fixed-length block and applying it to the pixel data of the blocks belonging to each fixed-length block within each group having substantially uniform characteristics related to image formation. When the edge portion and the flat portion are mixed in the image, the edge portion and the flat portion are grouped, so that the bit allocation is biased to the edge portion within one fixed-length block, and it is not reduced to the flat portion. . As a result, the distortion of the flat portion can be reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram showing the operation of an image encoding device embodying the present invention.

FIG. 2 is a diagram showing divided blocks and regions on a one-field screen of an image encoding device embodying the present invention.

FIG. 3 is a block diagram showing the operation of a conventional image encoding device.

FIG. 4 is a diagram showing a problem image generated by a conventional image encoding device.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 Blocking part 2 Area division part 3 Grouping part 4 Fixed-length blocking part 5 Orthogonal transformation part 6 Classification part 7 Bit allocation part 8 Quantization part 9 Encoding part 10 Block 11 One screen (one field or one frame) 12 area 13 background (forest, etc.) 14 person 15 edge part 16 flat part 17 boundary line 17 ′ boundary line 18 unevenness 18 ′ unevenness G1 group (areas 12a, b, c, e, f, g,
G2 group (configured from regions 12d, h, and l) 100 blocking unit 101 fixed-length blocking unit 102 orthogonal transform unit 103 classification unit 104 bit allocation unit 105 quantization unit 106 encoding unit

Claims (1)

(57) [Claims] 1. A method according to claim 1, wherein a predetermined number of adjacent pixels are grouped into one block, and one block is formed from image data of a plurality of pixels.
The screen image data is divided into a plurality of blocks of image data, and one fixed-length block is formed from a plurality of non-adjacent blocks. For each block belonging to one fixed-length block, all pixel data existing in that block In an image coding apparatus for orthogonally transforming to obtain an orthogonal transform coefficient, quantizing and encoding the same, a predetermined number of adjacent blocks are collectively set as a region, and a region dividing means for dividing a screen by the region; Based on the statistical results of the characteristics related to the image formation of the image data of the block to which it belongs, the areas where the results can be regarded as almost the same are included in this group as one group, and even in the areas not included, the results are almost the same. By repeatedly including the areas that can be considered as one group, several groups are And a grouping means for dividing the flop, the image coding apparatus, characterized in that each group formed by the grouping means, for generating a necessary and sufficient number of bits to allocate one fixed-length block.