JPH0795574A - Picture coder - Google Patents

Abstract

PURPOSE:To provide a picture coder in which the compression efficiency is improved when picture data are coded hierarchically and the deterioration in the picture quality is reduced. CONSTITUTION:When a quantization step width p1 of lower layer data is decided, the quantization step width adaptive to layer data D41, D42, D43, D44, D45 is decided by deciding a gain G multiplied with a quantization step width p0 of higher layer data adjacent the lower layer data while referring to a history of the result of selection of a gain G of the layer data higher than the lower layer data and compression coding data D51-D55 with less picture quality deterioration are obtained.

Description

Detailed Description of the Invention

[0001]

[Table of Contents] The present invention will be described in the following order. Field of Industrial Application Conventional Technology (FIGS. 9 and 10) Problem to be Solved by the Invention Means for Solving the Problem (FIGS. 4, 7 and 8) Action (FIGS. 4, 7 and 8) Embodiment (1) Principle of hierarchical coding (FIGS. 1 to 3) (2) Image coding apparatus of embodiments (FIGS. 4 to 6) (3) Selection of quantization step width (FIG. 7) (4) Selection of Quantization Step Width Based on History (FIG. 7) (5) Operation of Embodiment (FIG. 8) (6) Effect of Embodiment (7) Effect of Other Embodiment

[0002]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image coding apparatus,
For example, it is suitable for application to an image encoding device that divides predetermined image data into a plurality of image data having different resolutions and encodes the image data.

[0003]

2. Description of the Related Art Conventionally, as this type of image coding apparatus,
There is one in which input image data is hierarchically encoded using a hierarchical encoding method such as pyramid encoding. In this image encoding device, high-resolution input image data
As the hierarchical data, the second resolution data having a lower resolution than the first hierarchical data, the third hierarchical data having a lower resolution than the second resolution data, and so on are sequentially formed recursively. The plurality of hierarchical data are transmitted through one communication path or a transmission path that is a recording / reproducing path.

In addition, in the image decoding apparatus for decoding the plurality of hierarchical data, in addition to combining all of the plurality of image data, any one of the hierarchical data can be obtained depending on the resolution of the television monitor corresponding to each of them. One of them may be selected and decoded. As a result, by decoding only the desired hierarchical data from the plurality of hierarchical data, it is possible to obtain the desired image data with the minimum required transmission data amount.

Here, as shown in FIG. 9, in the image coding apparatus 1 which realizes, for example, four-layer coding as the hierarchical coding, thinning filters 2, 3, 4 for three stages are respectively provided.
And the interpolation filters 5, 6, and 7, and the input image data D
With respect to No. 1, the reduced image data D2, D3, D4 having a lower resolution are sequentially formed by the thinning filters 2, 3, 4 of each stage, and the reduced image data D2, D3, D4 are reduced by the interpolation filters 5, 6, 7 before reduction. Return to the resolution of.

Outputs D2 to D4 of the thinning filters 2 to 4
And the outputs D5 to D7 of the respective interpolation filters 5 to 7 are input to the difference circuits 8, 9 and 10, respectively, whereby difference data D8, D9 and D10 are generated. As a result, in the image encoding device 1, the amount of hierarchical data is reduced and the signal power is reduced. Here, this difference data D8
The area of each of D10 to D10 and the reduced image data D4 is 1,
The sizes are 1/4, 1/16 and 1/64.

The difference data D8 to D10 obtained from the respective difference circuits 8 to 10 and the reduced image data D4 obtained from the thinning filter 4 are the encoders 11, 12 and 1, respectively.
3 and 14 are used for encoding and compression processing, and as a result, the first, second, third and fourth hierarchical data D11, D1 having different resolutions from the respective encoders 11, 12, 13, 14 are obtained.
2, D13 and D14 are sent to the transmission line in a predetermined order.

The first to fourth hierarchical data D11 to D14 thus transmitted are decoded by the image decoding device 20 shown in FIG. That is, the first to fourth hierarchical data D11 to D14 are respectively decrypted by the decoders 21, 22, and
As a result, the fourth layer data D24 is output from the decoder 24.

The output of the decoder 23 is the fourth hierarchical data D obtained from the interpolation filter 26 in the adder circuit 29.
It is added with the interpolation data of 24, whereby the third hierarchical data D23 is restored. Similarly, the output of the decoder 22 is added to the interpolation data of the third hierarchical data D23 obtained from the interpolation filter 27 in the adder circuit 30, whereby the second hierarchical data D22 is restored. Further, the output of the decoder 21 is output by the adder circuit 31 to the interpolation filter 2
8 is added to the interpolated data of the second hierarchical data D22 to restore the first hierarchical data D21.

[0010]

However, in the image coding apparatus which realizes such a hierarchical coding method, since the input image data is divided into a plurality of hierarchical data for coding, only the hierarchical component data is inevitably obtained. There is a problem in that the amount of data increases and the compression efficiency decreases correspondingly as compared with a high efficiency coding method that does not use hierarchical coding. Further, when trying to improve the compression efficiency, there is a problem that the image quality is deteriorated by the quantizer applied between the respective hierarchical data.

The present invention has been made in consideration of the above points, and proposes an image coding apparatus which can improve compression efficiency and reduce image quality deterioration when hierarchically coding image data. Is.

[0012]

In order to solve such a problem, according to the present invention, a plurality of hierarchical data D32, D having a plurality of resolutions that sequentially recursively change the image data D31.
33, D34 and D35 and each hierarchical data D3
2, D33, D34, and D35 corresponding to each other, inter-layer data D41 in the block based on the quantization step width p0 of the upper layer data having a low resolution.
By determining the quantized values of D42, D43 and D44 and multiplying the gain G selected based on the distribution of the quantized values in this block by the quantization step width p0 of the upper layer data, In the image coding apparatus 40 for determining the quantization step width p1 of the lower layer data having a higher resolution, when determining the quantization step width p1 of the lower layer data, the quantization step of the upper layer data adjacent to the lower layer data is determined. The gain G multiplied by the width p0 is the gain G in the upper layer data than the lower layer data.
Make a decision by referring to the history of the selection results of.

Further, in the present invention, the image data D31
Is sequentially recursively divided into a plurality of hierarchical data D32, D33, D34 and D35 having different resolutions, and the upper hierarchical layer having a lower resolution for each block corresponding to each other among the hierarchical data D32, D33, D34 and D35. The quantization value of the inter-layer data D41, D42, D43 and D44 in the block is determined based on the quantization step width p0 of the data, and the gain G selected based on the distribution of the quantization value in this block is In the image coding device 40, which determines the quantization step width p1 of lower layer data having a higher resolution than the upper layer data by multiplying the quantization step width p0 of the upper layer data, the quantization step width p1 of the lower layer data. When determining, the gain G multiplied by the quantization step width p0 of the upper layer data adjacent to the lower layer data is set to the lower layer data. Data is determined with reference to the history of the selection results of the gain G in the upper layer data, and at this time, a different gain determination rule (Rule 1-
1 to Rule 1-4, Rule 2 and Rule 3) or (Rule 1 to Rule 3) are used.

[0014]

When determining the quantization step width p1 of the lower layer data, the gain G multiplied by the quantization step width p0 of the upper layer data adjacent to the lower layer data is multiplied by the gain G in the layer data higher than the lower layer data. If the determination is made by referring to the history of the selection results of the above, it is possible to determine the quantization step width adapted to each hierarchical data D41, D42, D43, D44 and D45, and thus compression coding with little deterioration in image quality is performed. Data D51 to D55 can be obtained.

If different gain determination rules (Rule 1-1 to Rule 1-4, Rule 2 and Rule 3) or (Rule 1 to Rule 3) are used for each layer, it is possible to use different gain determination rules. The quantization step width can be determined, and the image coding device 40 with high coding efficiency can be realized.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail below with reference to the drawings.

(1) Principle of Hierarchical Coding FIG. 1 shows, as a whole, the principle of hierarchical coding according to the present invention in which a still image such as a high definition television signal is hierarchically coded and compressed. In this hierarchical encoding, the upper layer data is created by a simple arithmetic average of the lower layer data, thereby realizing a hierarchical structure without an increase in the amount of information.
For decoding from the upper layer to the lower layer, the amount of information in the flat portion is reduced by adaptively controlling the division based on the activity of each block. Further, in the encoding of the differential signal performed for the lower layer, the efficiency is improved by switching the quantization characteristic for each block without additional code based on the activity of the upper layer.

That is, in the hierarchical structure of this hierarchical encoding,
First, the high-definition television signal to be input is defined as the lower layer, and the four pixels X1 to X4 in the small block of 2 lines × 2 pixels in this lower layer are expressed by the following equations.

[Equation 1] The arithmetic mean represented by is taken, and its value m is taken as the value of the upper hierarchy. In this lower hierarchy,

[Equation 2] As shown in, the difference value from the upper layer is prepared for three pixels, so that the hierarchical structure is configured with the same information amount as the original four pixel data.

On the other hand, when decoding the lower hierarchy, 3 pixels X1
~ X3 is the following formula

[Equation 3] As shown by, each difference value Δ is added to the average value m of the upper hierarchy.
Xi is added to obtain the decoded value E [Xi], and the remaining one pixel is

[Equation 4] As shown by, the decoded value E [X4] is determined by subtracting the three decoded values of the lower layer from the average value m of the upper layer. Here, E [] means a decoded value.

In this hierarchical coding, the resolution is quadrupled from upper layer to lower layer for each layer, but in the flat part, this division is prohibited to reduce redundancy. A flag for instructing the presence / absence of this division is prepared for each bit and block. The necessity of division in the lower layer is determined as local activity, for example, the maximum value of the difference data.

Here, as an example of hierarchical coding, HD of ITE is used.
FIG. 2 shows the adaptive division result when the standard image (Y signal) is used and five-layer coding is performed. The ratio of the number of pixels in each layer to the original number of pixels when the threshold value for the maximum difference data is changed is shown, and it can be seen that redundancy is reduced based on the spatial correlation. The reduction efficiency changes depending on the image, but if the threshold value for the maximum difference data is changed from 1 to 6, the average reduction rate becomes 28 to 69 [%].

Actually, the resolution of the upper layer is quadrupled to form the lower layer, and at that time, the difference data from the upper layer data is encoded in the lower layer, whereby the signal level width can be effectively reduced. FIG. 3 shows the case of five layers by the layer coding described above with reference to FIG. 2, but here, the layers are named from the lower layers to the first to fifth layers.

Compared to the 8-bit PCM data of the original image,
A reduction in signal level width can be seen. First with a large number of pixels
Since the ~ 4 layers are differential signals, a significant reduction can be achieved and efficiency is improved by subsequent quantization. As can be seen from the table of FIG. 3, the reduction efficiency has little dependence on the pattern, and is effective for all pictures.

By forming the upper layer with the average value of the lower layer, the error propagation is stopped in the block, and the lower layer is converted into the difference from the average value of the upper layer, so that the efficiency is also provided. You can In actual upper layer coding, there is a correlation in the activities between layers at the same spatial position, and by determining the quantization characteristic of the lower layer from the quantization result of the upper layer, adaptive quantization that does not require additional code Can be realized.

Practically, various HD standard images (8) are obtained by hierarchically encoding an image based on the above-described five-stage hierarchical structure and expressing it in multi-resolution, and performing adaptive division and adaptive quantization using the hierarchical structure. Y / PB / PR of bit is about 1 /
It can be compressed to 8. The additional code for each block prepared for adaptive division is run-length coded in each layer to improve compression efficiency. In this way, an image with sufficient image quality can be obtained in each layer, and a good image without visual deterioration can be obtained in the final lowest layer.

(2) Image Encoding Device of Embodiment In FIG. 4, reference numeral 40 denotes an image encoding device according to the present invention, and input image data D31 is input to the difference circuit 41 and the averaging circuit 42. As shown in FIG. 5, the averaging circuit 42 includes four pixels X1 (1) to X4 (1) of the input image data D31 which is the first layer data as the lowest layer to the pixel X1 (of the second layer data D32. 2) is generated. Similarly, the pixels X2 (2) to X4 (2) adjacent to the pixel X1 (2) of the second layer data D32 also have the first layer data D31.
It is generated by averaging 4 pixels.

The second layer data D32 is input to the difference circuit 43 and the averaging circuit 44. The averaging circuit 44 has a second
The third pixel data D is obtained by averaging four pixels of the hierarchical data D32.
33 is generated. For example, in the case shown in FIG. 5, the pixels X1 (2) to X4 (2) of the second layer data D32 generate the pixels X1 (3) of the third layer data D33, and
Pixels X2 (3) to X4 (3) adjacent to the pixel X1 (3)
Is also generated by averaging four pixels of the second layer data D32.

The third layer data D33 is input to the difference circuit 45 and the averaging circuit 46. As in the case described above, the averaging circuit 46 averages four pixels of the third layer data D33 to generate the fourth layer data D34 of pixels X1 (4) to X4 (4) as shown in FIG. The fourth layer data D34 is input to the difference circuit 47 and the averaging circuit 48. The averaging circuit 48 generates the fifth layer data D35 which is the highest layer by averaging four pixels of the fourth layer data D34. As shown in FIG. 5, the fourth layer data D34
The pixel X1 (5) of the fifth hierarchical data D35 is generated by averaging the four pixels X1 (4) to X4 (4).

Therefore, the first to fifth layer data D31 to D3
As for the block size of 5, if the block size of the first layer data D31 which is the lowest layer is 1 × 1, the second layer data D32 is 1/2 × 1/2 and the third layer data D33 is 1/3.
4 × 1/4, the fourth hierarchical data D34 is 1/8 × 1/8, and the fifth hierarchical data D35, which is the highest hierarchical data, is 1/16 × 1/16.

For example, when upper layer data is generated by averaging four pixels of spatially corresponding lower layer data, if the upper layer data is M and the lower layer pixel values are a, b, c, d, the transmission pixels are It will be good with 4 pixels. That is, using M, a, b, c, d, the following equation,

[Equation 5] The non-transferred pixel d can be easily restored on the decoder side by the arithmetic expression represented by

A schematic diagram of the relationship between the layers is shown in FIG. 6 for an example of four layers. Here, each layer data is 4
It is generated by pixel averaging, and all the data can be restored by the arithmetic expression shown in the equation (5) without transmitting the shaded data in the figure. As a result, in the image coding device 40, the number of pixels to be coded by the subsequent coder can be reduced, and thus, even when the coding is performed after disassembling into a plurality of hierarchical images, it is possible to avoid a decrease in compression efficiency. It is done like this.

In the image coding apparatus 40, the encoder 49 compresses and codes the fifth layer data D35 to generate fifth layer compression coded data D55. Further, in the image encoding device 40, inter-layer difference data D44, D is obtained by performing a difference operation between adjacent layers for each of the above five respective layer data D31 to D35.
43, D42, D41 are generated.

That is, in the image coding device 40,
First, the fourth layer data D34 is input to the difference circuit 47, and the fifth layer compression encoded data D55 is restored by the decoder 50 and input as the restored data D36. As a result, the difference circuit 47 generates inter-layer difference data D44 between the fourth layer data D34 and the fifth layer data D35, and outputs this to the encoder 51. The encoder 51 compression-codes the inter-layer difference data D44 to generate fourth layer compression-coded data D54.

Next, in the image encoding device 40, the third layer data D33 is input to the difference circuit 45, and the fourth layer compression encoded data D54 is restored by the decoder 52 to obtain the fourth layer data D34. Restored data D similar to
Enter 37. As a result, the difference circuit 45 generates inter-layer difference data D43 between the third layer data D33 and the restored data D37 (that is, the fourth layer data D34), and outputs this to the encoder 53. The encoder 53 compression-codes the inter-layer difference data D43 to generate third-layer compression-coded data D53.

Similarly, in the image coding device 40,
The second layer data D32 is input to the difference circuit 43, and the third layer compression encoded data D53 is decompressed by the decoder 54 to restore the same decompressed data D33 as the third layer data D33.
Enter 38. As a result, the difference circuit 43 generates inter-layer difference data D42 between the second layer data D32 and the restored data D38 (that is, the third layer data D33), and outputs this to the encoder 55. The encoder 55 compression-codes the inter-layer difference data D42 to generate second-layer compression-coded data D52.

Finally, the image coding device 40 inputs the first layer data D31 to the difference circuit 41 and the second layer data D31.
The layer compression-encoded data D52 is decompressed by the decoder 56, and decompressed data D39 similar to the second layer data D32 is input. As a result, the difference circuit 41 causes the first layer data D3
1 and the restored data D39 (that is, the second layer data D3
Inter-layer difference data D41 with 2) is generated and output to the encoder 57. The encoder 57 uses the inter-layer difference data D4.
1 is compression-encoded to generate first layer compression-encoded data D51.

As described above, in the image encoding device 40, the fifth layer compression encoded data D55, the fourth layer compression encoded data D54, the third layer compression encoded data D53,
The second layer compression coded data D52 and the first layer compression coded data D51 are sequentially generated in this order.

(3) Selection of Quantization Step Width Here, the encoders 49, 51, 53, 55 and 57 each have a quantizer. In the image encoding device 40, when the lower layer data area corresponding to the upper layer data is defined as "block", the characteristics of the data change in the block are grasped by the activity of the inter-layer difference data D41 to D44 in this block. , The characteristic of the quantizer is determined based on this data characteristic. Here, the activity is a correlation value that can be obtained from the maximum value, average value, sum of absolute values, standard deviation, sum of n powers, etc. of the inter-tier difference data D41 to D44 in a predetermined block.

In the case of the embodiment, a 2-bit quantizer is used as the quantizer, and the difference value is +12 in this quantizer.
FIG. 7 shows the quantization characteristics when two-bit quantization is performed on the difference data between layers in the range of 8 to -128. In this way, the difference value is quantized into 0 to 3. Further, in the case of the embodiment, since upper layer data is generated by averaging 4 × 2 × 2 pixels for each layer data, there are 4 pixels in the lower layer of each block.

Here, as a method of determining the quantization characteristic of each quantizer, first, the inter-layer difference data is 2-bit quantized according to the quantization step width determined in the upper layer. At this time, one of the quantized values 0 to 3 shown in FIG. 7 is generated. Here, since the distribution of the quantized values of the four pixels in the block represents the activity in the block, the quantization step width of the next layer is determined based on the quantized value distribution of the four pixels. Thus, by selecting the quantization step width based on the intra-block quantized value distribution, an additional code indicating the type of quantizer becomes unnecessary.

As a result, in the image coding device 40,
It is possible to improve the compression efficiency of the encoders 49, 51, 53, 55, 57 and reduce image quality deterioration during the compression encoding process. Next, the rule for determining the quantization step width in the embodiment will be described. First, the quantized values 0 to 3 are classified into sections A and B as shown in FIG. That is, when the quantized value is 1 or 2, it is set as the section A, and when the quantized value is 0 or 3, it is set as the section B.

Here, as a characteristic of the quantizer for efficiently forming a high quality image, a coarse quantizer having a large quantization step width is used for a block having a high activity, and a low activity is used for the block. Considering that it is necessary to use a quantizer having a narrow quantization step width in the block, the following rules are set.

That is, in the quantizer, when the quantization step width of the upper layer is p0 and the quantization step width of the lower layer is p1, rule 1) when the quantized values of 4 pixels all belong to section B ,
p1 = 2 × p0 Rule 2) When the quantized values of 4 pixels belong to the intervals A and B, p1 = p0 Rule 3) When the quantized values of 4 pixels all belong to the interval A,
The quantization step width p1 of the lower layer is determined based on p1 = p0 / 2.

Here, rule 1 corresponds to the case where the intra-block activity is large, and in this case, the quantization step width of the next lower layer is increased to fulfill the function of suppressing the quantization distortion. Rule 2 can be considered as a state of intra-block activity in many cases, but it is generally considered that the absolute value of the data in section B is not large due to the spatial correlation, so the quantization step width of the upper layer is maintained. Fulfill the function of Furthermore, rule 3 corresponds to the case where the intra-block activity is small, and in this case, the quantization step width of the next lower layer is made small, and the function of suppressing the image quality deterioration in the flat part is fulfilled.

As described above, in the image coding apparatus 40, the quantization step width of the lower layer is determined according to the intra-block activity of the upper layer.

(4) Selection of Quantization Step Width Based on History Further, in the image coding device 40, in addition to determining the quantization step width in accordance with Rule 1 to Rule 3 as described above, at this time, The record of the determination result of the quantization step width in the upper layer than the layer to be determined, that is, the selection history of the quantization step width in the upper layer is reflected in the layer for which the current quantization step width is determined. There is.

The above-mentioned rules 1 to 3 are expressed as p1 = G × p0 using the gain G, and determine the gain G according to the combination of the quantized values of 4 pixels. Here, in the quantizer of the embodiment, the gain G more suited to the activity is determined based on the decision history of the hierarchy higher than the hierarchy to be determined and the rules 1 to 3, and based on this gain G It is designed to determine the quantization step width. It should be noted that the history of determination of the quantization step width in upper gymnastics can be said in other words as the history of the selection result of the gain G.

For the sake of explanation, the intra-block quantized value patterns are classified as follows. Pattern 1) When the quantized values of 4 pixels all belong to section B. Pattern 2) When the quantized values of 4 pixels belong to section A and section B. Pattern 3) When the quantized values of 4 pixels all belong to section A. Further, the frequency of each pattern in the decision history of the hierarchy higher than the hierarchy for which the quantized value is determined is defined as follows. N1) The frequency of pattern 1 in the upper hierarchy determination history. N2) The frequency of pattern 2 in the upper hierarchy determination history. N3) The frequency of pattern 3 in the upper hierarchy determination history.

In the quantizer, using the patterns 1 to 3 and N1 to N3, the above rule 1 is classified into the following rules 1-1 to 1-4 in detail, and this rule 1-1 is used.
~ Determine the quantization step width based on the gain G obtained by rules 1-4. Rule 1-1) In the case of pattern 1 and N3 = 0,
G = 2 Rule 1-2) In the case of pattern 1 and N1 = 0,
G = 1.5 Rule 1-3) If pattern 1 and N1> TH0 and N3> TH1, G = 1.0. Where TH0 and TH
1 is a threshold value of the pattern occurrence frequency, and the threshold values TH0 and TH1 are determined according to the layer number (first layer to fifth layer). Rule 1-4) In the case of pattern 1 and the upper hierarchy determination history is other than the above, G = 2.0

The reason why rule 1 is further classified into rules 1-1 to 1-4 is that rule 1 has a large gain (G = 2) with respect to the quantization step width in the upper layer.
This is because the gain G may oscillate and the quantization step width may oscillate in the determination over a plurality of layers. That is, if the current quantization step width is determined based on only the activity of the upper layer immediately before the layer for which the quantization step width is to be determined, as in Rules 1 to 3, the gain G = 2, The gain G = 1/2 appears to intersect with each other, and at this time, the gain G oscillates, and an appropriate quantization step width cannot be selected.

Therefore, in the quantizer of the embodiment, the gain G in the upper hierarchy is based on the rules 1-1 to 1-4.
The gain G is converged in accordance with the characteristics of the hierarchical image by considering the selection history of No. 1, so that the image quality deterioration due to the oscillation of the gain G can be avoided.

That is, rule 1-1 is a case where a high block activity is recognized even in the selection history, and at this time, the quantizer of this hierarchy has a large gain (G =
2) is given, which means to determine the quantization step width. Rule 1-2 is a case where the block activity is not high in the selection history, which means that the quantizer gradually lowers the gain G at this time. Furthermore, Rule 1-3 is a case where both a large gain G and a small gain G appear in the selection history,
At this time, the quantizer may hold the previous value of the gain G because the gain G may oscillate. Furthermore, rules 1-4 mean that the general processing of pattern 1 is performed.

Thus, in the quantizer of the embodiment, the gain G of the quantization step width is determined based on the rules 1-1 to 1-4, the rules 2 and 3, and the determined gain G is adjacent to the gain G. By deciding the quantization step width of the current layer by multiplying it by the quantization step width of the upper layer, the oscillation of the quantization step width due to the gain G can be avoided in advance. It is possible to further reduce image deterioration at that time.

(5) Operation of the Embodiment With the above configuration, the image encoding device 40 sequentially generates the first to nth hierarchical compression code data according to the processing procedure as shown in FIG. 8 (of the embodiment). Case n = 5). That is, the image coding apparatus 40 enters from the step SP1 and inputs n-1 to the layer counter I in step SP2, assuming n layers.

The image coding apparatus 40 uses the following step SP.
In 3, the averaging circuits 42, 44, 46 and 48 generate the hierarchical data D31 to D35 for n layers, and the process proceeds to step SP4. Here, the image coding apparatus 40 sets an initial value of the quantization step width which is an attribute of the highest layer. The image encoding device 40, in the following step 5,
Encoding and decoding processing of the highest layer data D35 is executed. Incidentally, at this time, the image coding apparatus 40 operates in the step S.
The top layer data D35 is not quantized by the initial value of the quantization step width initialized in P4,
The initial value of the quantization step width is set as an initial value for determining the quantization step width in the lower layer.

Next, the image coding device 40 proceeds to step SP6.
Then, the difference circuit 47, 45, 43 or 41 first performs the inter-layer difference calculation, and the inter-layer difference data D44, D43, D42 or D41 generated at this time is compared with the quantization step width of the upper layer. Quantize by.
Next, the image coding apparatus 40 judges the distribution of the quantized value in the block in step SP7, and judges in accordance with the above-mentioned rules 1-1 to 1-4, rule 2 and rule 3 in step SP8. Then, the quantization step width is determined based on the determination result, and the quantization step width is transmitted to the lower layer.

The image coding device 40 uses the following step SP.
In 9, the encoding and decoding of the inter-layer difference data D44, D43, D42 or D41 is executed by using the quantization step width determined in step SP8. The image encoding device 40 decrements the hierarchical counter I in step SP10, and determines whether the hierarchical counter I is 0 in the following step SP11.

When an affirmative result is obtained here, this means that the processing of all the layers is completed, and at this time, the image coding apparatus 40 moves to step SP12 and ends the processing procedure. On the other hand, if a negative result is obtained in step SP11, the image coding apparatus 40 proceeds to step S.
Returning to P5, the above step S for the next lower layer
The processing from P5 to step SP10 is repeated.

(6) Effects of the Embodiments According to the above configuration, the gain G for determining the quantization step width p1 of the lower layer is determined with reference to the selection history of the gain G in the upper layer. By doing so, an appropriate quantization step width can be obtained for each hierarchical data, and the image encoding device 40 with reduced image quality deterioration.
Can be realized.

(7) Other Embodiments In the above-described embodiment, the quantizers of all the layers have the gain G of the quantization step width based on the rules 1-1 to 1-4, rule 2 and rule 3. However, the present invention is not limited to this, and it is not always necessary to determine the quantization step width based on the rules 1-1 to 1-4, rule 2 and rule 3 over all layers. Instead, the gain determination rule applied for each layer may be changed.

That is, in the hierarchical encoding, considering that the appearance of the image quality deterioration and the encoding efficiency differ between the layers due to the difference in the image size between the layers, the same gain determination is not necessarily applied to all the layers. Since it is not considered necessary to apply rules, for example, in the lower hierarchy,
While the above-mentioned rules 1-1 to 1-4, rules 2 and 3 are applied to determine the quantization step width, rules 1 to 3 may be applied in the higher layer. good. By doing so, the quantization step width can be determined according to the appearance of the image quality, and an image coding apparatus with high coding efficiency can be obtained.

In the above-described embodiment, the gain determination rules for reflecting the determination history of the quantization step width in the upper layer on the quantization step width in the lower layer are rule 1 to rule 1.
The case of using the rule 3 and the rules 1-1 to 1-4 has been described, but the present invention is not limited to this, and various gain determination rules can be applied. It is only necessary that the determination history of the quantization step width can be reflected in the quantization step width of the lower layer.

In this case, the gain G for reflecting the determination history of the quantization step width in the upper layer to the quantization step width in the lower layer can be expressed as W _i (H, p0) as a function. The quantization step width p1 is obtained by using the quantization step width of the adjacent upper layer as p1.
= W _i (H, p0) × p0.

[0064]

As described above, according to the present invention, when the quantization step width of the lower layer data is determined, the gain multiplied by the quantization step width of the upper layer data adjacent to the lower layer data is set to the lower level. By making a decision by referring to the history of gain selection results in higher hierarchical data than hierarchical data, it is possible to obtain an appropriate quantization step width for each hierarchical data, and to reduce the image degradation An encoding device can be realized.

[Brief description of drawings]

FIG. 1 is a schematic diagram for explaining hierarchical data generated by an image encoding device according to the present invention.

FIG. 2 is a chart showing an adaptive division result in an HD standard image.

FIG. 3 is a chart showing standard deviations of signal levels of respective layers in an HD standard image.

FIG. 4 is a block diagram showing a circuit configuration of an embodiment of an image encoding device according to the present invention.

FIG. 5 is a schematic diagram for explaining a generation operation of hierarchical data.

FIG. 6 is a schematic diagram for explaining a hierarchical structure of hierarchical data.

FIG. 7 is a schematic diagram showing characteristics of a quantizer according to an embodiment.

[Fig. 8] Fig. 8 is a flowchart for explaining an operation of hierarchical encoding.

FIG. 9 is a block diagram showing a conventional image encoding device.

FIG. 10 is a block diagram showing a conventional image decoding apparatus.

[Explanation of symbols]

40 ... Image coding device, 41, 43, 45, 47 ...
Difference circuit, 42, 44, 46, 48 ... Averaging circuit, 4
9, 51, 53, 55, 57 ... Encoder, 50, 52,
54, 56 ... Decoder, D31 to D35 ... Hierarchical data, D41 to D44 ... Difference data between layers, D51 to D
55 ... Hierarchical compression and decoding data.

[Procedure amendment]

[Submission date] November 24, 1994

[Procedure Amendment 1]

[Document name to be amended] Statement

[Correction target item name] Full text

[Correction method] Change

[Correction content]

[Document name] Statement

Image coding apparatus

[Claims]

Detailed Description of the Invention

[0001]

[Table of Contents] The present invention will be described in the following order. Industrial Application Conventional Technology (FIGS. 11 and 12) Problems to be Solved by the Invention Means for Solving the Problems (FIGS. 4, 8 and 9) Operation (FIGS. 4, 8 and 9) Embodiment (1) Principle of hierarchical coding (FIGS. 1 to 3) (2) Image coding apparatus of embodiments (FIGS. 4 to 8) (3) Selection of quantization step width (FIG. 9) (4) Selection of Quantization Step Width Based on History (FIG. 9) (5) Operation of Embodiment (FIG. 10) (6) Effect of Embodiment (7) Effect of Other Embodiment

[0002]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image coding apparatus,
For example, it is suitable for application to an image encoding device that divides predetermined image data into a plurality of image data having different resolutions and encodes the image data.

[0003]

2. Description of the Related Art Conventionally, as this type of image coding apparatus,
There is one in which input image data is hierarchically encoded using a hierarchical encoding method such as pyramid encoding. In this image encoding device, high-resolution input image data
As the hierarchical data, the second resolution data having a lower resolution than the first hierarchical data, the third hierarchical data having a lower resolution than the second resolution data, and so on are sequentially formed recursively. , The plurality of hierarchical data are transmitted through a transmission line such as a communication line or a recording / reproducing route.

Further, in the image decoding apparatus for decoding the plurality of hierarchical data, all of the plurality of hierarchical data may be decoded, and any one of the hierarchical data may be decoded depending on the resolution of the television monitor corresponding to each of them. One of them may be selected and decrypted. As a result, by decoding only the desired hierarchical data from the plurality of hierarchical data, it is possible to obtain the desired image data with the minimum required transmission data amount.

Here, as shown in FIG. 11, in the image coding apparatus 1 which realizes, for example, four-layer coding as the hierarchical coding, thinning filters 2, 3,
4 and interpolation filters 5, 6, and 7, which form reduced image data D2, D3, and D4 of low resolution sequentially by the thinning filters 2, 3, and 4 of each stage for the input image data D1 and the interpolation filter. The reduced image data D2, D3, and D4 are returned to the resolution before reduction by 5, 6, and 7.

Outputs D2 to D4 of the thinning filters 2 to 4
And the outputs D5 to D7 of the respective interpolation filters 5 to 7 are input to the difference circuits 8, 9 and 10, respectively, whereby difference data D8, D9 and D10 are generated. As a result, in the image encoding device 1, the amount of hierarchical data is reduced and the signal power is reduced. Here, this difference data D8
The area of each of D10 to D10 and the reduced image data D4 is 1,
The sizes are 1/4, 1/16, and 1/64.

The difference data D8 to D10 obtained from the respective difference circuits 8 to 10 and the reduced image data D4 obtained from the thinning filter 4 are the encoders 11, 12 and 1, respectively.
3 and 14 are used for encoding and compression processing, and as a result, the first, second, third and fourth hierarchical data D11, D1 having different resolutions from the respective encoders 11, 12, 13, 14 are obtained.
2, D13 and D14 are sent to the transmission line in a predetermined order.

The first to fourth hierarchical data D11 to D14 thus transmitted are decoded by the image decoding device 20 shown in FIG. That is, the first to fourth hierarchical data D11 to D14 are respectively decrypted by the decoders 21, 22, and
As a result, the fourth layer data D24 is output from the decoder 24.

The output of the decoder 23 is the fourth hierarchical data D obtained from the interpolation filter 26 in the adder circuit 29.
It is added with the interpolation data of 24, whereby the third hierarchical data D23 is restored. Similarly, the output of the decoder 22 is added to the interpolation data of the third hierarchical data D23 obtained from the interpolation filter 27 in the adder circuit 30, whereby the second hierarchical data D22 is restored. Further, the output of the decoder 21 is output by the adder circuit 31 to the interpolation filter 2
8 is added to the interpolated data of the second hierarchical data D22 to restore the first hierarchical data D21.

[0010]

However, in the image coding apparatus which realizes such a hierarchical coding method, since the input image data is divided into a plurality of hierarchical data for coding, only the hierarchical component data is inevitably obtained. There is a problem in that the amount of data increases and the compression efficiency decreases correspondingly as compared with a high efficiency coding method that does not use hierarchical coding. Further, when trying to improve the compression efficiency, there is a problem that the image quality is deteriorated by the quantizer applied between the respective hierarchical data.

The present invention has been made in consideration of the above points, and proposes an image coding apparatus which can improve compression efficiency and reduce image quality deterioration when hierarchically coding image data. Is.

[0012]

In order to solve such a problem, according to the present invention, a plurality of hierarchical data D32, D having a plurality of resolutions that sequentially recursively change the image data D31.
33, D34 and D35 and each hierarchical data D3
2, D33, D34, and D35 corresponding to each other, inter-layer data D41 in the block based on the quantization step width p0 of the upper layer data having a low resolution.
By determining the quantized values of D42, D43, and D44 and multiplying the gain G selected based on the distribution of the quantized values in this block by the quantization step width p0 of the upper layer data, In the image coding device 40 for determining the quantization step width p1 of lower layer data having a higher resolution than that, when determining the quantization step width p1 of the lower layer data, the quantization of the upper layer data adjacent to the lower layer data is performed. The gain G multiplied by the step width p0 is set to the gain G in the higher layer data than the lower layer data.
Make a decision by referring to the history of the selection results of.

Further, in the present invention, the image data D31
Is sequentially recursively divided into a plurality of hierarchical data D32, D33, D34 and D35 having different resolutions, and the upper hierarchical layer having a lower resolution for each block corresponding to each other among the hierarchical data D32, D33, D34 and D35. The quantization value of the inter-layer data D41, D42, D43, and D44 in the block is determined based on the quantization step width p0 of the data, and the gain G selected based on the distribution of the quantization value in this block is In the image coding device 40, which determines the quantization step width p1 of the lower layer data having a higher resolution than the upper layer data by multiplying the quantization step width p0 of the upper layer data, the quantization step width p1 of the lower layer data. When determining, the gain G by which the quantization step width p0 of the upper layer data adjacent to the lower layer data is multiplied In this case different gain determination rules for each layer with the selection result of the history of the gain G in the upper hierarchy data is determined with reference than the data (rules 1
1 to Rule 1-4, Rule 2 and Rule 3) or (Rule 1 to Rule 3) are used.

[0014]

When determining the quantization step width p1 of the lower layer data, the gain G multiplied by the quantization step width p0 of the upper layer data adjacent to the lower layer data is multiplied by the gain G in the layer data higher than the lower layer data. If the determination is made by referring to the history of the selection result, it is possible to determine the quantization step width adapted to each layer data D41, D42, D43, D44 and D45, and thus compression encoding with little deterioration in image quality is performed. Data D51 to D55 can be obtained.

If different gain determination rules (Rule 1-1 to Rule 1-4, Rule 2 and Rule 3) or (Rule 1 to Rule 3) are used for each layer, it is possible to use different gain determination rules. The quantization step width can be determined, and the image coding device 40 with high coding efficiency can be realized.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail below with reference to the drawings.

(1) Principle of Hierarchical Coding FIG. 1 shows, as a whole, the principle of hierarchical coding according to the present invention in which a still image such as a high definition television signal is hierarchically coded and compressed. In this layer coding, the upper layer data is created by a simple arithmetic average of the lower layer data, the lower layer data to be transmitted is reduced, and the layer structure without increasing the information amount is realized. For decoding from the upper layer to the lower layer, the amount of information in the flat portion is reduced by adaptively controlling the division based on the activity of each block. Further, in the encoding of the differential signal performed for the lower layer, the efficiency is improved by switching the quantization characteristic for each block without additional code based on the activity of the upper layer.

That is, in the hierarchical structure of this hierarchical encoding,
First, the high-definition television signal to be input is defined as the lower layer, and the four pixels X1 to X4 in the small block of 2 lines × 2 pixels in this lower layer are expressed by the following equations.

[Equation 1] The arithmetic mean represented by is taken, and its value m is taken as the value of the upper hierarchy. In this lower hierarchy,

[Equation 2] As shown in, the difference value from the upper layer is prepared for three pixels, so that the hierarchical structure is configured with the same information amount as the original four pixel data.

On the other hand, when decoding the lower layer, 3 pixels X1
~ X3 is the following formula

[Equation 3] As shown by, each difference value Δ is added to the average value m of the upper hierarchy.
Xi is added to obtain the decoded value E [Xi], and the remaining one pixel is

[Equation 4] As shown by, the decoded value E [X4] is determined by subtracting the three decoded values of the lower layer from the average value m of the upper layer. Here, E [] means a decoded value.

In this hierarchical coding, the resolution is quadrupled from upper layer to lower layer for each layer, but in the flat part, this division is prohibited to reduce redundancy. A flag for instructing the presence / absence of this division is prepared for each bit and block. The necessity of division in the lower layer is determined as local activity, for example, the maximum value of the difference data.

Here, as an example of hierarchical coding, HD of ITE is used.
FIG. 2 shows the adaptive division result when the standard image (Y signal) is used and five-layer coding is performed. The ratio of the number of pixels in each layer to the original number of pixels when the threshold value for the maximum difference data is changed is shown, and it can be seen that redundancy is reduced based on the spatial correlation. Although the reduction efficiency changes depending on the image, if the threshold value for the maximum difference data is changed to 1 to 6, the average reduction rate becomes 28 to 69 [%].

Actually, the resolution of the upper layer is quadrupled to form the lower layer, and at that time, the difference data from the upper layer data is encoded in the lower layer, whereby the signal level width can be effectively reduced. FIG. 3 shows the case of five layers by the layer coding described above with reference to FIG. 2, but here, the layers are named from the lower layers to the first to fifth layers.

Compared to the 8-bit PCM data of the original image,
A reduction in signal level width can be seen. First with a large number of pixels
Since the ~ 4 layers are differential signals, a significant reduction can be achieved and efficiency is improved by subsequent quantization. As can be seen from the table of FIG. 3, the reduction efficiency has little dependence on the pattern, and is effective for all pictures.

By forming the upper layer with the average value of the lower layer, the error propagation is stopped in the block, and the lower layer is converted into the difference from the average value of the upper layer, so that the efficiency is also provided. You can Actually, in the upper layer coding, there is a correlation in the activities between layers at the same spatial position, and by determining the quantization characteristic of the lower layer from the quantization result of the upper layer, the receiver side can be used for dequantization. It is possible to realize an adaptive quantizer that does not need to transmit quantization information (however, except for initial values).

Practically, various HD standard images (8) are obtained by hierarchically encoding an image based on the above-described five-stage hierarchical structure and expressing it in multi-resolution, and performing adaptive division and adaptive quantization using the hierarchical structure. Bit / Y / PB / PR) is about 1 /
It can be compressed to 8. The additional code for each block prepared for adaptive division is run-length coded in each layer to improve compression efficiency. In this way, an image with sufficient image quality can be obtained in each layer, and a good image without visual deterioration can be obtained in the final lowest layer.

(2) Image Encoding Device of Embodiment In FIG. 4, reference numeral 40 denotes an image encoding device according to the present invention, and input image data D31 is input to the difference circuit 41 and the averaging circuit 42. As shown in FIG. 5, the averaging circuit 42 includes four pixels X1 (1) to X4 (1) of the input image data D31, which is the first hierarchy data as the lowest hierarchy, to the pixel X1 (of the second hierarchy data D32. 2) is generated. Similarly, the pixels X2 (2) to X4 (2) adjacent to the pixel X1 (2) of the second layer data D32 also have the first layer data D31.
Is generated by averaging four pixels.

The second layer data D32 is input to the difference circuit 43 and the averaging circuit 44. The averaging circuit 44 has a second
Third layer data D is obtained by averaging four pixels of layer data D32.
33 is generated. For example, in the case shown in FIG. 5, the pixels X1 (2) to X4 (2) of the second layer data D32 generate the pixels X1 (3) of the third layer data D33, and
Pixels X2 (3) to X4 (3) adjacent to the pixel X1 (3)
Is similarly generated by averaging four pixels of the second layer data D32.

The third layer data D33 is input to the difference circuit 45 and the averaging circuit 46. As in the case described above, the averaging circuit 46 averages four pixels of the third layer data D33 to generate the fourth layer data D34 including pixels X1 (4) to X4 (4) as shown in FIG. The fourth layer data D34 is input to the difference circuit 47 and the averaging circuit 48. The averaging circuit 48 generates the fifth layer data D35, which is the highest layer, by averaging four pixels of the fourth layer data D34. As shown in FIG. 5, by averaging the four pixels X1 (4) to X4 (4) of the fourth layer data D34, the pixel X1 (5) of the fifth layer data D35 is generated.

Therefore, the first to fifth layer data D31 to D3
As for the block size of 5, when the block size of the first layer data D31 which is the lowest layer is 1 × 1, the second layer data D32 is 1/2 × 1/2 and the third layer data D33.
Is 1/4 × 1/4, and the fourth layer data D34 is 1/8 × 1
/ 8, and the fifth layer data D35, which is the highest layer data, is 1/16 × 1/16.

For example, when the upper layer data is generated by averaging four pixels of the spatially corresponding lower layer data, if the upper layer data is M and the lower layer pixel values are a, b, c, d, the transmission pixels are , The upper layer data M and the lower layer pixels a, b, and c may be left as they are. That is, using M, a, b, c, d, the following equation,

[Equation 5] The non-transferred pixel d can be easily restored on the decoder side by the arithmetic expression represented by

A schematic diagram of the relationship between the layers is shown in FIG. 6 for an example of four layers. Here, each layer data is 4 of the lower layer.
It is generated by pixel averaging, and all the data can be restored by the arithmetic expression shown in the equation (5) without transmitting the shaded data in the figure. As a result, in the image coding device 40, the number of pixels to be coded by the subsequent coder can be reduced, and thus, even when the coding is performed after disassembling into a plurality of hierarchical images, it is possible to avoid a decrease in compression efficiency. It is done like this.

In the image encoding device 40, the encoder 49 compression-encodes the fifth layer data D35 to generate fifth layer compression encoded data D55. Further, in the image encoding device 40, inter-layer difference data D44, D is obtained by performing a difference operation between adjacent layers for each of the above five respective layer data D31 to D35.
43, D42, D41 are generated.

That is, in the image coding device 40,
First, the fourth layer data D34 is input to the difference circuit 47, and the fifth layer compression encoded data D55 is restored by the decoder 50 and input as the restored data D36. As a result, the difference circuit 47 generates inter-layer difference data D44 between the fourth layer data D34 and the fifth layer data D35, and outputs this to the encoder 51. The encoder 51 compression-codes the inter-layer difference data D44 to generate fourth layer compression-coded data D54.

Next, in the image encoding device 40, the third layer data D33 is input to the difference circuit 45, and the fourth layer compression encoded data D54 is restored by the decoder 52 to obtain the fourth layer data D34. Restored data D similar to
Enter 37. As a result, the difference circuit 45 generates inter-layer difference data D43 between the third layer data D33 and the restored data D37 (that is, the fourth layer data D34), and outputs this to the encoder 53. The encoder 53 compression-codes the inter-layer difference data D43 to generate third-layer compression-coded data D53.

Similarly, in the image coding device 40,
The second layer data D32 is input to the difference circuit 43, and the third layer compression encoded data D53 is decompressed by the decoder 54 to restore the same decompressed data D33 as the third layer data D33.
Enter 38. As a result, the difference circuit 43 generates inter-layer difference data D42 between the second layer data D32 and the restored data D38 (that is, the third layer data D33), and outputs this to the encoder 55. The encoder 55 compression-codes the inter-layer difference data D42 to generate second-layer compression-coded data D52.

Finally, the image coding device 40 inputs the first layer data D31 to the difference circuit 41 and the second layer data D31.
The layer compression-encoded data D52 is decompressed by the decoder 56, and decompressed data D39 similar to the second layer data D32 is input. As a result, the difference circuit 41 causes the first layer data D3
1 and the restored data D39 (that is, the second layer data D3
Inter-layer difference data D41 with 2) is generated and output to the encoder 57. The encoder 57 uses the inter-layer difference data D4.
1 is compression-encoded to generate first layer compression-encoded data D51.

As described above, in the image encoding device 40, the fifth layer compression encoded data D55, the fourth layer compression encoded data D54, the third layer compression encoded data D53,
The second layer compression coded data D52 and the first layer compression coded data D51 are sequentially generated in this order.

Here, each of the decoders 52, 54 and 56 receives the compression coded data D54, D53 or D52 to be decoded from the corresponding encoder 51, 53 or 55, and the corresponding encoders 51, 53 respectively. Alternatively, the compression encoded data D54, D53 or D52 is decoded by receiving the quantization information E0, E1 or E2 used in 55.
Further, each of the decoders 52, 54 and 56 is the restored data D36, D37 from the decoder 50, 52 or 56 in the next lower layer.
Or by receiving the data in D38, the hierarchical data D before the difference
Make 34, D33 or D32.

In practice, each decoder 52, 54, 56 is shown in FIG.
It is configured as shown in. Here, the decoder 52 will be described for simplification. The decoder 52 is the decoding circuit 5
2A receives the fourth layer compression encoded data D54 and the quantization information E0 used when generating the fourth layer compression encoding data D54, and decodes the fourth layer compression encoding data D54. From the result decoding circuit 52A, for example, as shown in FIG.
X1 (4) -X1 (5) and X2 (4) -X1 shown in
Output values of (5) and X3 (4) -X1 (5) are obtained.
This output value is reconstructed in the adder circuit 52B by the restored data D.
By adding 36, X1 (4), X2 (4),
The output value of X3 (4) is obtained. Difference value generation circuit 52C
Is X1 (4), X2 (4), X3 (4) and X1 (5)
Is used to perform the calculation based on the equation (5) to generate the non-transmission pixel X4 (4). Accordingly, the following synthesis circuit 52
From D, the fourth layer data X1 (4), X2 before the difference
(4), X3 (4), and X4 (4) are generated and provided to the difference circuit 45.

The encoders 51, 53, 55, 57 receive the quantized information E0, E1, E2 or E3 output from the encoders 49, 51, 53, 55 of the upper layers adjacent to each other, and the quantized information concerned. Each difference data is quantized based on E0, E1, E2, and E3. In addition, the image encoding device 4
0 is the quantized information E0, E1, E sent from the respective encoders 49, 51, 53, 55 to the adjacent lower layers.
2 and E3 are also given to the quantization width control circuit 61. The quantization width control circuit 61 outputs history signals S1, S2, S3, and S4, which represent the history of the quantization step width selection in the upper hierarchy up to that time, to the encoders 51, based on the quantization information E0 to E3.
53, 55, 57.

Here, the encoders 51, 53, 55 and 57 are constructed as shown in FIG. In FIG. 8, the encoders 53 and 55 will be described for simplification. That is, the inter-layer difference data D43 sent to the encoder 53 is the quantizer 5
3A, the quantizer 53A receives the quantization information (quantization step width) received from the encoder 51 of the upper layer.
The inter-layer difference data D43 is quantized based on E1.
The quantized value obtained as a result is given to the distribution state determination circuit 53D together with the code word assignment circuit 53B, the distribution state is determined by the distribution state determination circuit 53D, and the determination result is the quantization width selection circuit 53C. Given to.

The quantization width selection circuit 53C multiplies the quantization information (quantization step width) E1 by the gain based on the determination result from the distribution state determination circuit 53D and the history signal S2 to obtain a new quantization step width. It is generated and sent to the encoder 55 and the quantization width control circuit 61 in the adjacent lower layer as the quantization information E2.

Similarly, the encoder 55 quantizes the inter-layer difference data D42 based on the quantization information (quantization step width) E2 received from the encoder 53, and distributes the distribution state of the resulting quantized values. The state determination circuit 55D makes a determination, and supplies the determination result to the quantization width selection circuit 55C.
The quantization width selection circuit 55C generates a new quantization step width by multiplying the quantization information (quantization step width) E2 by the gain based on the determination result from the distribution state determination circuit 55D and the history signal S3, This is quantized information E3
Is sent to the encoder 57 and the quantization width control circuit 61 of the adjacent lower layer.

As a result, in the image coding device 40,
It is possible to further improve the compression efficiency of the encoders 51, 53, 55, 57 and further reduce image quality deterioration during the compression encoding process.

(3) Selection of Quantization Step Width In the image coding device 40, the lower layer data area corresponding to the upper layer data is defined as "block".
The characteristic of the data change in the block is grasped by the activity of the inter-layer difference data D41 to D44 in the block, and the characteristic of the quantizer is determined based on this data characteristic.

In the case of the embodiment, the quantizers 53A, 55A,
.. is a 2-bit quantizer, and FIG. 9 shows the quantization characteristic when the 2-bit quantizer is used to quantize inter-layer difference data having a difference value in the range of +128 to -128. In this way, the difference value is quantized into 0 to 3. Further, in the case of the embodiment, each hierarchical data is 2 × 2 = 4.
Since upper layer data is generated by pixel averaging,
There are 4 pixels in the lower hierarchy of each block.

Here, as a method of determining the quantization characteristic of each quantizer, first, the difference data between layers is 2-bit quantized according to the quantization step width determined in the upper layer. At this time, any of the quantized values 0 to 3 shown in FIG. 9 is generated. Since the distribution of the quantized values of the four pixels in the block represents the activity in the block, the quantization step width of the next layer is determined based on the quantized value distribution of the four pixels. Thus, by selecting the quantization step width based on the intra-block quantized value distribution, an additional code indicating the type of quantizer becomes unnecessary.

As a result, in the image coding device 40,
It is possible to improve the compression efficiency of the encoders 49, 51, 53, 55, 57 and reduce image quality deterioration during the compression encoding process. Next, the rule for determining the quantization step width in the embodiment will be described. First, the quantized values 0 to 3 are classified into sections A and B as shown in FIG. That is, when the quantized value is 1 or 2, it is set as the section A, and when the quantized value is 0 or 3, it is set as the section B.

Here, as a characteristic of the quantizer for efficiently forming a high quality image, a coarse quantizer having a large quantization step width is used for a block having a high activity, and a low activity is used for the block. Considering that it is necessary to use a quantizer having a narrow quantization step width in the block, the following rules are set.

That is, in the quantizer, when the quantization step width of the upper layer is p0 and the quantization step width of the lower layer is p1, rule 1) when the quantized values of 4 pixels all belong to the section B. ,
p1 = 2 × p0 Rule 2) When the quantized values of 4 pixels belong to section A and section B, p1 = p0 Rule 3) When all the quantized values of 4 pixels belong to section A,
The quantization step width p1 of the lower layer is determined based on p1 = p0 / 2.

Here, rule 1 corresponds to the case where the intra-block activity is large, and in this case, the quantization step width of the next lower layer is increased to fulfill the function of suppressing the quantization distortion. Rule 2 can be considered as a state of intra-block activity in many cases, but it is generally considered that the absolute value of the data in section B is not large due to the spatial correlation, so the quantization step width of the upper layer is maintained. Fulfill the function of Furthermore, rule 3 corresponds to the case where the intra-block activity is small, and in this case, the quantization step width of the next lower layer is made small, and the function of suppressing the image quality deterioration in the flat part is fulfilled.

As described above, in the image coding apparatus 40, the quantization step width of the lower layer is determined according to the intra-block activity of the upper layer.

(4) Selection of Quantization Step Width Based on History Further, in the image coding device 40, in addition to determining the quantization step width in accordance with Rule 1 to Rule 3 as described above, at this time, The record of the determination result of the quantization step width in the upper layer than the layer to be determined, that is, the selection history of the quantization step width in the upper layer is reflected in the layer for which the current quantization step width is determined. There is.

In the above-mentioned rules 1 to 3, p1 = G × p0 is expressed by using the gain G, and the gain G is determined according to the combination of the quantized values of 4 pixels. Here, in the image encoding device 40 of the embodiment, the decision history (that is, the history signals S1 to S of the hierarchy higher than the hierarchy to be determined).
4) and the rules 1 to 3 are used to determine the gain G more suited to the activity, and the gain G is multiplied by the quantization step width used in the current layer to obtain a new quantization step width. It is designed to decide. The quantization step width determination history in the upper layer can be said to be the history of the gain G selection result.

For the sake of explanation, the intra-block quantized value patterns are classified as follows. Pattern 1) When all the quantized values of 4 pixels belong to the section B. Pattern 2) When the quantized values of 4 pixels belong to the section A and the section B. Pattern 3) When all the quantized values of 4 pixels belong to the section A. Further, the frequency of each pattern in the decision history of the hierarchy higher than the hierarchy for which the quantized value is determined is defined as follows. N1) The frequency of pattern 1 in the upper hierarchy determination history. N2) The frequency of pattern 2 in the upper hierarchy determination history. N3) The frequency of pattern 3 in the upper hierarchy determination history.

In the quantization width selection circuits 53C, 55C, ..., Using the patterns 1 to 3 and N1 to N3, the above rule 1 is changed to the following rules 1-1 to 1-4.
Then, the quantization step width is determined based on the gain G obtained by the rules 1-1 to 1-4. Rule 1-1) In the case of pattern 1 and N3 = 0,
G = 2 Rule 1-2) In the case of pattern 1 and N1 = 0,
G = 1.5 Rule 1-3) If pattern 1 and N1> TH0 and N3> TH1, G = 1.0. Where TH0 and T
H1 is a threshold value of the pattern occurrence frequency, and the threshold values TH0 and TH1 are layer numbers (first layer to fifth layer).
According to. Rule 1-4) In the case of pattern 1 and the upper hierarchy determination history is other than the above, G = 2.0

The reason why rule 1 is further classified into rules 1-1 to 1-4 is that rule 1 has a large gain (G = 2) with respect to the quantization step width in the upper layer.
This is because the gain G may oscillate and the quantization step width may oscillate in the determination over a plurality of layers. That is, if the current quantization step width is determined based on only the activity of the upper layer immediately before the layer for which the quantization step width is to be determined, as in Rules 1 to 3, the gain G = 2, The gain G = 1/2 appears in an intersecting manner, and at this time, the gain G oscillates and an appropriate quantization step width cannot be selected.

Therefore, in the image coding apparatus 40 of the embodiment, by considering the selection history of the gain G in the higher hierarchy based on the rules 1-1 to 1-4, the characteristics of the hierarchy image can be adjusted. The gain G is converged so that the image quality deterioration due to the oscillation of the gain G is avoided.

That is, rule 1-1 is a case where a high block activity is recognized even in the selection history, and at this time, the quantizer of this hierarchy has a large gain (G =
2) is given, which means to determine the quantization step width. Rule 1-2 is a case where the block activity is not high in the selection history, which means that the quantizer gradually lowers the gain G at this time. Furthermore, Rule 1-3 is a case where both a large gain G and a small gain G appear in the selection history,
At this time, the quantizer may hold the previous value of the gain G because the gain G may oscillate. Furthermore, rules 1-4 mean that the general processing of pattern 1 is performed.

Thus, in the image coding apparatus 40 of the embodiment, the gain G of the quantization step width is determined based on the rules 1-1 to 1-4, the rules 2 and 3, and the determined gain G is determined. By deciding the quantization step width of the current layer by multiplying by the quantization step width of the adjacent upper layer, oscillation of the quantization step width due to the gain G can be avoided in advance. It is possible to further reduce the image deterioration at the time of conversion.

(5) Operation of the Embodiment With the above configuration, the image coding apparatus 40 sequentially generates the first to nth hierarchical compression code data according to the processing procedure as shown in FIG. 10 (of the embodiment). Case n = 5). That is, the image encoding device 40 enters from the step SP1 and at the step SP2, the layer counter I assuming n layers.
Enter n-1 into

The image coding apparatus 40 uses the following step SP.
In 3, the averaging circuits 42, 44, 46 and 48 generate the hierarchical data D31 to D35 for n layers, and the process proceeds to step SP4. Here, the image coding apparatus 40 sets an initial value of the quantization step width which is an attribute of the highest layer. The image encoding device 40, in the following step 5,
Encoding and decoding processing of the highest layer data D35 is executed. Incidentally, at this time, the image coding apparatus 40 operates in the step S.
The top layer data D35 is not quantized by the initial value of the quantization step width initialized in P4,
The initial value of the quantization step width is set as an initial value for determining the quantization step width in the lower layer.

Next, the image encoding device 40 proceeds to step SP6.
Then, the difference circuit 47, 45, 43 or 41 first performs the inter-layer difference calculation, and the inter-layer difference data D44, D43, D42 or D41 generated at this time is compared with the quantization step width of the upper layer. Quantize by.
Next, the image coding apparatus 40 judges the distribution of the quantized value in the block in step SP7, and judges in accordance with the above-mentioned rules 1-1 to 1-4, rule 2 and rule 3 in step SP8. Then, the quantization step width is determined based on the determination result, and the quantization step width is transmitted to the lower layer.

The image coding apparatus 40 uses the following step SP.
In 9, the encoding and decoding of the inter-layer difference data D44, D43, D42 or D41 is executed by using the quantization step width determined in step SP8. The image encoding device 40 decrements the hierarchical counter I in step SP10, and determines whether the hierarchical counter I is 0 in the following step SP11.

When an affirmative result is obtained here, this means that the processing of all the layers is completed, and at this time, the image coding apparatus 40 moves to step SP12 and ends the processing procedure. On the other hand, if a negative result is obtained in step SP11, the image coding apparatus 40 proceeds to step S.
Returning to P5, the above step S for the next lower layer
The processing from P5 to step SP10 is repeated.

(6) Effects of the Embodiments According to the above configuration, the gain G for determining the quantization step width p1 of the lower layer is determined with reference to the selection history of the gain G in the upper layer. By doing so, an appropriate quantization step width can be obtained for each hierarchical data, and the image encoding device 40 with reduced image quality deterioration.
Can be realized.

(7) Other Embodiments In the above-described embodiment, the gain G of the quantization step width for the quantizers of all layers is set to the rules 1-1 to 1-4, rule 2 and rule 3. However, the present invention is not limited to this, and the quantization step width is determined based on the rules 1-1 to 1-4, rule 2 and rule 3 over all layers. There is no need, and the gain determination rule to be applied may be changed for each layer.

In other words, in the hierarchical coding, considering that the image size deterioration and the coding efficiency differ between layers due to the difference in image size between layers, the same gain determination is not necessarily performed over all layers. Since it is not considered necessary to apply rules, for example, in the lower hierarchy,
While the above-mentioned rules 1-1 to 1-4, rules 2 and 3 are applied to determine the quantization step width, rules 1 to 3 may be applied in the higher layer. good. By doing so, the quantization step width can be determined according to the appearance of the image quality, and an image coding apparatus with high coding efficiency can be obtained.

Further, in the above-described embodiment, the gain determination rules for reflecting the quantization step width determination history in the upper layer on the quantization step width in the lower layer are rule 1 to rule 1.
The case of using the rule 3 and the rules 1-1 to 1-4 has been described, but the present invention is not limited to this, and various gain determination rules can be applied. It is only necessary that the determination history of the quantization step width can be reflected in the quantization step width of the lower layer.

In this case, the gain G for reflecting the determination history of the quantization step width in the upper layer on the quantization step width in the lower layer can be expressed as a function W _i (H, p0). The quantization step width p1 is p1 by using the quantization step width of the adjacent upper layer as p0.
= W _i (H, p0) × p0.

[0071]

As described above, according to the present invention, when the quantization step width of the lower layer data is determined, the gain multiplied by the quantization step width of the upper layer data adjacent to the lower layer data is set to the lower level. By making a decision by referring to the history of gain selection results in higher hierarchical data than hierarchical data, it is possible to obtain an appropriate quantization step width for each hierarchical data, and to reduce the image degradation An encoding device can be realized.

[Brief description of drawings]

FIG. 1 is a schematic diagram for explaining hierarchical data generated by an image encoding device according to the present invention.

FIG. 2 is a chart showing an adaptive division result in an HD standard image.

FIG. 3 is a chart showing standard deviations of signal levels of respective layers in an HD standard image.

FIG. 4 is a block diagram showing a circuit configuration of an embodiment of an image encoding device according to the present invention.

FIG. 5 is a schematic diagram for explaining a generation operation of hierarchical data.

FIG. 6 is a schematic diagram for explaining a hierarchical structure of hierarchical data.

FIG. 7 is a block diagram showing the configuration of a decoder.

FIG. 8 is a block diagram showing a configuration of an encoder.

FIG. 9 is a schematic diagram showing characteristics of a quantizer according to an example.

[Fig. 10] Fig. 10 is a flow chart for explaining an operation of hierarchical encoding.

FIG. 11 is a block diagram showing a conventional image encoding device.

FIG. 12 is a block diagram showing a conventional image decoding apparatus.

[Explanation of Codes] 40 ... Image coding device, 41, 43, 45, 47 ...
Difference circuit, 42, 44, 46, 48 ... Averaging circuit, 4
9, 51, 53, 55, 57 ... Encoder, 50, 52,
54, 56 ... Decoder, D31 to D35 ... Hierarchical data, D41 to D44 ... Difference data between layers, D51 to D
55 ... Hierarchical compression and decoding data.

[Procedure Amendment 2]

[Document name to be corrected] Drawing

[Correction target item name] All drawings

[Correction method] Change

[Correction content]

[Figure 1]

[Figure 9]

[Fig. 2]

[Figure 3]

[Figure 4]

[Figure 5]

[Figure 6]

[Fig. 12]

[Figure 7]

[Figure 10]

[Figure 8]

FIG. 11

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Office reference number FI technical display location H04N 1/41 B

Claims (2)

[Claims] 1. A quantizing step width of upper layer data having a lower resolution for each block corresponding to each other among the plurality of hierarchical data having a plurality of different resolutions recursively sequentially recursively dividing the image data. Based on, to determine the quantization value of the inter-layer data in the block, the gain selected based on the distribution of the quantization value in the block, by multiplying the quantization step width of the upper layer data, In the image coding apparatus for determining the quantization step width of lower layer data having a higher resolution than the upper layer data, when determining the quantization step width of the lower layer data, the upper layer data adjacent to the lower layer data is determined. Determine the gain to be multiplied by the quantization step width by referring to the history of gain selection results in higher layer data than the lower layer data. An image encoding device characterized by the following. 2. The image data is sequentially recursively divided into a plurality of hierarchical data having a plurality of different resolutions, and for each block corresponding to each other among the hierarchical data, the quantization step width of the upper hierarchical data having a low resolution Based on, to determine the quantization value of the inter-layer data in the block, the gain selected based on the distribution of the quantization value in the block, by multiplying the quantization step width of the upper layer data, In the image coding apparatus for determining the quantization step width of the lower layer data having a higher resolution than the layer data, when determining the quantization step width of the lower layer data, of the upper layer data adjacent to the lower layer data Determine the gain to be multiplied by the quantization step width by referring to the history of gain selection results in higher layer data than the lower layer data. In addition, at this time, an image coding apparatus is characterized in that a different gain determination rule is used for each layer.