JP3324548B2 - Motion compensation prediction decision circuit - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention particularly relates to a motion compensation prediction decision circuit for selecting a prediction mode based on AE (sum of absolute differences).

[0002]

2. Description of the Related Art Conventionally, an international standard I of moving picture compression
TU-TH. In TM5 (test model 5) of H.262 / MPEG2, motion compensation treats all prediction modes as candidates as shown in FIG. 6 and performs AE (Absolute).
Error: the prediction mode with the smallest absolute difference value) is selected.

[0003] Please refer to FIG. First, the process up to the calculation of the AE will be described. The MB average 501 is equal to the predicted image signal Int.
The average of the macroblock (256 samples) of the ra input is output. The AE device 503 is connected to the MB average 501,
The Intra macroblock average is subtracted from the predicted image signal Intra, the absolute value is calculated, the result is accumulated for each macroblock, and IP (AE of Intra prediction) is output.
FP (Forward prediction AE), BP (Backw
AE) and IPP (Inter Polat)
AE of ion prediction has the same configuration. The (F + B) / 2 unit 502 has a predicted image signal For
Ward input and predicted image signal Backward input 1
Average value for each sample (= Inter Polation)
Prediction) is calculated and output to the AE unit 506.

Thus, four AEs are required. The comparator 507 is connected to the AE units 503, 504, 505, and 506, finds the minimum AE among IP, FP, BP, and IPP, and outputs a prediction mode corresponding to the minimum AE.

[0005]

However, the prior art has the following problems. In the conventional prediction mode determination circuit described above, DCT (Discrete C
The optimal mode selection is performed on the “Osine Transform” coefficient. In this selection, since the information amount of the motion vector to be encoded is not included in the mode selection condition as a variable, the entire encoded information is There was a problem that the optimum mode could not be determined. That is,
In the mode selection in which all the prediction modes are treated as candidates, the code amount of DCT is selected so as to be small.
There is a case where the code amount of the motion vector information increases, and therefore, there is no guarantee that an optimal selection is made for the entire coded information.

[0006] The present invention has been made in view of such a problem, and an object thereof is to provide an FP (Forwa).
RD prediction AE), BP (Backward prediction A
E), IPP (interpolation prediction A)
If the value of E (Absolute Error) increases, the motion vector information tends to increase. If the AE value is equal to or greater than a certain threshold, the number of motion vectors to be encoded is the largest.・ P (Int
It is an object of the present invention to provide a configuration in which an optimal mode can be determined for the entire code information by performing mode selection by excluding (er / Polation) prediction from selection targets.

[0007]

Means for Solving the Problems In order to solve the above problems, the present invention has the following constitution. The gist of the present invention is a motion compensation prediction determination circuit for selecting a prediction mode of an image based on a sum of absolute differences,
FP calculating means for calculating and outputting the sum of absolute differences (FP) of the prediction, and the sum of absolute differences (B
BP calculating means for calculating and outputting P);
IPP calculating means for calculating and outputting the sum of absolute differences (IPP) in the Position prediction, and the sum of absolute differences (FP) in the forward prediction and the sum of absolute differences (BP) in the backward prediction are input and input. If the smallest sum of the absolute differences of the differences (FP, BP) is larger than the threshold,
Prediction mode corresponding to sum of absolute differences (IPP) of Position prediction (Inter Poll prediction)
And a mode selection restricting means for excluding from the target of the prediction mode selection. The gist of the present invention resides in the motion compensation prediction determining circuit according to claim 1, wherein the mode selection restricting means receives a sum of absolute differences (IPP) of the InterPollation prediction. The gist of the invention according to claim 3 is that the motion compensation prediction determination circuit according to claim 1 or 2, further comprising IP calculation means for calculating and outputting the sum of absolute differences (IP) of the intra prediction. Exist. The gist of the invention according to claim 4 is that the In
The sum of absolute differences (IP) of the tra prediction and the Forwa
rd prediction sum of absolute difference (FP) and the Backwa
The sum of absolute differences (BP) of rd prediction and the Inter
4. The method according to claim 1, further comprising a first mode selection unit that selects and outputs a prediction mode corresponding to a minimum sum of absolute differences between the IPPs of the Position prediction. In the motion compensation prediction determining circuit described above. The gist of the invention described in claim 5 is that the Intra
A prediction mode corresponding to the smallest sum of absolute differences of prediction (IP), the sum of absolute differences of forward prediction (FP), and the sum of absolute differences of backward prediction (BP) is selected. The motion compensation prediction determining circuit according to any one of claims 1 to 4, further comprising a second mode selecting means for outputting the result. The gist of the invention according to claim 6 is that, when the smallest sum of absolute differences among the input sums of absolute differences is larger than a threshold, the mode selection restricting means outputs the output of the second mode selecting means. By selecting, the Inter Pol
The motion compensation prediction determination circuit according to claim 4 or 5, wherein a prediction mode (Inter Pollation prediction) corresponding to the sum of absolute differences (IPP) of the prediction of the prediction is excluded from the target of the prediction mode selection. .
The gist of the invention according to claim 7 is that, when the smallest sum of absolute differences among the input sums of difference absolute values is smaller than a threshold, the mode selection restricting means outputs the output of the first mode selection means. 7. The motion compensation prediction determining circuit according to claim 6, wherein The gist of the invention according to claim 8 is that the mode selection restricting means selects a minimum sum of absolute differences among the input sums of absolute differences, and compares the minimum sum of absolute differences with a threshold.
Comparing means for outputting the result, and a selector for selecting and outputting one of the output of the first mode selecting means and the output of the second mode selecting means according to the output of the comparing means The motion compensation prediction determining circuit according to claim 7, wherein The gist of the present invention resides in a motion compensation prediction determining circuit according to any one of claims 1 to 8, wherein the mode selection restricting means is constituted by a logic circuit. According to a tenth aspect of the present invention, there is provided the motion compensation prediction determining circuit according to any one of the first to ninth aspects, wherein the threshold is variable. The gist of the present invention resides in a computer base including the motion compensation prediction determining circuit according to any one of claims 1 to 10. The gist of the present invention resides in an image processing apparatus provided with the motion compensation prediction determining circuit according to any one of claims 1 to 10. The gist of the invention according to claim 13 is a motion compensation prediction determination method for selecting an image prediction mode based on a sum of absolute differences,
Calculate the sum of absolute differences (FP) of the forward prediction,
The difference absolute value sum (BP) of the Backward prediction is calculated, the difference absolute value sum (IPP) of the Inter Position prediction is calculated, and the difference absolute value sum (FP) of the Forward prediction and the difference absolute value sum (FP) of the Backward prediction are calculated. BP), when the smallest sum of absolute differences is larger than the threshold, the Inter Polati
There is a motion compensation prediction determination method for excluding a prediction mode (Inter Position prediction) corresponding to the sum of absolute differences (IPP) of on prediction from selection of a prediction mode. The gist of the invention according to claim 14 is a motion compensation prediction determining method for selecting an image prediction mode based on a sum of absolute differences, wherein a sum of absolute differences (FP) of forward prediction is calculated, and Sum of absolute differences (BP)
Is calculated, and the sum of absolute differences (IPP) of the Inter Poll prediction is calculated, and the sum of absolute differences (FP) of the Forward prediction, the sum of the absolute differences of the Backward prediction (BP), and the difference absolute value of the Inter Poll prediction are calculated. If the smallest sum of absolute differences is greater than the threshold value of the sum (IPP),
There is a motion compensation prediction determination method for excluding a prediction mode (Inter Position prediction) corresponding to the sum of absolute differences (IPP) of the Position prediction from the target of the prediction mode selection.

[0008]

FIG. 1 is a block diagram showing a motion compensation prediction determining circuit according to a first embodiment of the present invention. As shown in FIG. 1, in the motion compensation prediction determining circuit according to the first embodiment, the values of FP and BP are larger than a set threshold value (the threshold level can be adjusted from the outside). In this case, the IP prediction is removed from the selection targets, the remaining prediction modes are input to the comparator 108, the minimum AE is obtained by the comparator 108, and the prediction mode corresponding to the minimum AE is selected. Thereby, the image quality can be improved. On the other hand, when the values of FP and BP are smaller than the set threshold values, the processing of TM5 is performed.

Hereinafter, a first embodiment of the present invention will be described in detail with reference to the drawings.

Please refer to FIG. First, the process up to the calculation of the AE will be described. MB average 101 is equal to predicted image signal Int.
The average of the macroblock (256 samples) of the ra input is output. The AE device 103 is connected to the MB average 101,
The Intra macroblock average is subtracted from the predicted image signal Intra, the absolute value is calculated, the result is accumulated for each macroblock, and IP (AE of Intra prediction) is output.
FP (Forward prediction AE), BP (Backw
AE) and IPP (Inter Polat)
AE of ion prediction has the same configuration. The (F + B) / 2 unit 102 has a predicted image signal For
Ward input and predicted image signal Backward input 1
Average value for each sample (= Inter Polation)
Prediction) is calculated and output to the AE unit 106.

Thus, four AEs are obtained. The comparator 107 is connected to the AE units 103, 104, 105, and 106, finds the minimum AE among IP, FP, BP, and IPP, and outputs a prediction mode corresponding to the minimum AE.
The comparator 108 is connected to the AE units 103, 104, and 105, finds the smallest AE among IP, FP, and BP, and outputs a prediction mode corresponding to the smallest AE. MIN device 10
Reference numeral 9 is connected to the AE devices 104 and 105, and the threshold is externally input as a reference value. MIN device 109
Outputs a control signal SEL for selecting the output of the comparator 108 when both FP and BP are larger than the threshold. However, when FP and BP are both smaller than the threshold, the control signal SE for selecting the output of the comparator 107 is set.
L is output. The selector 110 includes comparators 107 and 108
And the control signal S output from the MIN unit 109
Either of the prediction modes given from the comparators 107 and 108 is output according to EL.

FIG. 2 is a block diagram showing the configuration of the MIN unit 109 shown in FIG. In FIG. 2, the MIN device 1
Reference numeral 09 includes a comparator 201 and a comparator 202. FP and BP are input to the comparator 201 and the comparator 2
01 compares these two input values and outputs the smaller one. The comparator 202 is connected to the comparator 201, compares the value set in the threshold with the value output from the comparator 201, and compares the threshold value with the value of the comparator 2
When the value given from the comparator 201 is large, the control signal SEL for selecting the output of the comparator 108 (FIG. 1) is output. When the value given from the comparator 201 is smaller than the threshold value, the comparator 107 (FIG. The control signal SEL for selecting the output of 1) is output. Next, the operation of the motion compensation prediction determination circuit shown in FIG. 1 will be described with reference to the drawings. The prediction determination block will be described assuming that a macroblock (256 samples) is used as a processing unit. First, IP,
The process up to the calculation of FP, BP, and IPP will be described.

The motion compensation prediction decision circuit shown in FIG.
Intra, Forward, Backward, Fo
The average value of rward and Backward for each sample is input as a prediction candidate. Here, only the method of obtaining the IP will be described, but the same applies to FP, BP, and IPP. First, the Intra and the average value of the Intra macroblock obtained by the MB average 101 are given to the AE unit 103, and the sum of absolute differences (= Intr
AE) IP of a prediction is calculated. Similarly, AE values, IP, FP, BP, and IPP of a total of four candidates are obtained.
However, regarding the subtraction processing inside the AE unit, only IP is M
The difference is that the subtraction with the B average is performed, but the subtraction process is performed with the Intra in other respects.

Please refer to FIG. FIG. 3 is a timing chart of the motion compensation prediction determination circuit shown in FIG.
As described above, the AE values of the four candidates obtained by the AE units 103, 104, 105, and 106 are IP, FP, BP,
The IPP is input to the comparators 107 and 108 and the MIN unit 109, respectively. At this moment, the threshold is assumed to be a fixed value.

In the first half of the timing chart of FIG.
Since both P and BP are smaller than the threshold value, M
The IN unit 109 supplies a control signal for selecting the comparator 107 to the selector 110, and the selector 110 outputs an Inter Patio having the minimum AE similarly to the TM5.
The n prediction is output as Mode. However, in the latter half, since both FP and BP are larger than the threshold value, the MIN unit 109 supplies a control signal for selecting the comparator 108 to the comparator 110 irrespective of the minimum IPP.
The smallest FP excluding the IPP, that is, the forward prediction, is selected as the mode from the comparator 110.

Note that the threshold is 1080I, 720
To cope with various conditions such as an image format such as P, MI
This is input to the N unit 109.

Next, a second embodiment of the present invention will be described in detail with reference to the drawings. FIG. 4 shows a motion compensation prediction determining circuit according to the second embodiment of the present invention. Referring to FIG. 4, only items different from the first embodiment of the present invention shown in FIG. 1 and the second embodiment of the present invention shown in FIG. 4 will be described.

First, until the calculation of the AE, the IP (Intr
a AE for prediction), FP (AE for forward prediction),
BP (backward prediction AE), IPP (Int
ERPolation prediction AE) is output in the same manner as in the first embodiment of the present invention shown in FIG. And MI
Compared to the first embodiment of the present invention shown in FIG. 1, the output from the AE unit 406 is newly connected to the N unit 409 in addition to the outputs from the AE units 404 and 405.

FIG. 5 is a block diagram showing a configuration of the MIN1 unit 409 shown in FIG. In FIG. 5, the MIN unit 409 includes a comparator 301, a comparator 302, and a comparator 303. FP and BP are input to the comparator 301, and the comparator 301 compares the two input values and outputs the smaller one. The output of the comparator 301 and the IPP are input to the comparator 302. The comparator 302 compares the two input values and outputs the smaller one. The comparator 303 is connected to the comparator 302 and compares the value set in the threshold with the value output from the comparator 302. When the value given from the comparator 302 is larger than the threshold value, the comparator 303 408 (FIG. 4), and outputs a control signal SEL to select the output of the comparator 30 from the threshold value.
When the value given from 1 is small, the control signal SEL for selecting the output of the comparator 407 (FIG. 4) is output. This has a new effect that the IP prediction is not selected even when the IPP is large.

Since the motion compensation prediction determining circuit according to these embodiments is configured as described above, the following effects can be obtained.

When the values of FP and BP (and IPP in the second embodiment) are larger than the threshold, the prediction mode IPP having a large code amount of the motion vector information is used.
Is excluded from the targets of mode selection, it is possible to select an optimal mode not only in DCT coefficients but also in motion vector information.

In the present embodiment, the present invention is not limited to this, and can be applied to a mode suitable for applying the present invention.

Further, the number, position, shape, etc. of the above-mentioned constituent members are not limited to the above-mentioned embodiment, but can be set to a suitable number, position, shape, etc. for carrying out the present invention.

In each of the drawings, the same components are denoted by the same reference numerals.

[0025]

Since the present invention is configured as described above, the following effects can be obtained. In addition to the motion compensation prediction determination method of TM5, when the value of AE is large, IP (In)
By not selecting a prediction mode, it is possible to determine an optimal mode for the entire code information. The reason is that if the value of AE is large,
This is because there is a circuit that excludes the I / P prediction mode signal, which has a large number of motion vectors to be handled, from the selection target because the information amount of the motion vector tends to be large.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a motion compensation prediction determination circuit according to a first embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of a MIN unit 109 shown in FIG.

FIG. 3 is a timing chart of the motion compensation prediction determination circuit shown in FIG.

FIG. 4 is a block diagram illustrating a motion compensation prediction determination circuit according to a second embodiment of this invention.

FIG. 5 is a block diagram showing a configuration of a MIN unit 409 shown in FIG.

FIG. 6 is a block diagram showing a conventional electric circuit for performing motion compensation.

[Explanation of symbols]

 101 MB average 102 (F + B) / 2 unit 103, 104, 105, 106 AE unit 107, 108 Comparator 109 MIN unit 110 Selector 201, 202 Comparator 301, 302, 303 Comparator 401 MB average 402 (F + B) / 2 403, 404, 405, 406 AE unit 407, 408 Comparator 409 MIN unit 410 Selector 501 MB average 502 (F + B) / 2 unit 503, 504, 505, 506 AE unit 507 Comparator

Claims (14)

(57) [Claims] 1. A motion compensation prediction determining circuit for selecting a prediction mode of an image based on a sum of absolute differences, FP calculating means for calculating and outputting a sum of absolute differences (FP) of Forward prediction, BP calculating means for calculating and outputting the sum of absolute differences (BP) of Backward prediction, and the sum of absolute differences (I
IPP calculating means for calculating and outputting the sum of absolute differences (FP) of the forward prediction and the sum of absolute differences (BP) of the backward prediction, and inputting the sum of absolute differences (FP) , BP) is larger than the threshold, the prediction mode (Inter P) corresponding to the sum of absolute differences (IPP) of the Inter Poll prediction.
motion prediction prediction determination circuit, comprising: a mode selection restricting unit that excludes the prediction prediction from the prediction mode selection. 2. The method according to claim 1, wherein the mode selection restricting means includes:
sum of absolute difference (IPP) of er Position prediction
2. The motion compensation prediction determining circuit according to claim 1, wherein the following is input. 3. Intra prediction sum of absolute differences (IP)
The motion compensation prediction determination circuit according to claim 1, further comprising an IP calculation unit that calculates and outputs the calculation result. 4. The sum of absolute differences (I
P) and the sum of absolute differences (F
P) and the sum of absolute differences (B
P) and a first mode selecting means for selecting and outputting a prediction mode corresponding to a minimum sum of absolute differences (IPP) of the Inter Position prediction. 4. The motion compensation prediction determination circuit according to any one of claims 1 to 3. 5. The sum of absolute differences (I
P) and the sum of absolute differences (F
P) and the sum of absolute differences (B
The motion compensation prediction determination circuit according to any one of claims 1 to 4, further comprising a second mode selection means for selecting and outputting a prediction mode corresponding to the smallest sum of absolute differences among P). 6. The mode selection limiting means selects an output of the second mode selection means when a minimum sum of absolute differences among the input sums of absolute differences is larger than a threshold. The Inter Pola
The motion compensation prediction determination circuit according to claim 4, wherein a prediction mode (Inter Position prediction) corresponding to a sum of absolute differences (IPP) of the tion prediction is excluded from prediction mode selection targets. 7. The mode selection restricting means selects an output of the first mode selecting means when a minimum difference absolute value sum of the input difference absolute value sums is smaller than a threshold. The motion compensation prediction determination circuit according to claim 6, wherein 8. The mode selection restricting means selects a minimum sum of absolute differences among the input sums of absolute differences, compares the minimum sum of absolute differences with a threshold, and outputs the result. And a selector for selecting and outputting one of an output of the first mode selecting means and an output of the second mode selecting means in accordance with an output of the comparing means. Item 7. The motion compensation prediction determination circuit according to Item 7. 9. The motion compensation prediction determining circuit according to claim 1, wherein said mode selection restricting means is constituted by a logic circuit. 10. The motion compensation prediction determination circuit according to claim 1, wherein the threshold is variable. 11. A computer base comprising the motion compensation prediction determining circuit according to claim 1. 12. An image processing apparatus comprising the motion compensation prediction determination circuit according to claim 1. 13. A motion compensation prediction determining method for selecting an image prediction mode based on a sum of absolute differences, comprising: calculating a sum of absolute differences (FP) of forward prediction;
The difference absolute value sum (BP) of the Backward prediction is calculated, the difference absolute value sum (IPP) of the Inter Position prediction is calculated, and the difference absolute value sum (FP) of the forward prediction and the difference absolute value sum ( BP), the smallest difference absolute value sum is larger than the threshold, the prediction mode (Inter P) corresponding to the difference absolute value sum (IPP) of the Inter Position prediction.
motion prediction), which excludes the prediction from the prediction mode selection. 14. A motion compensation prediction determining method for selecting an image prediction mode based on a difference absolute value sum, comprising: calculating a difference absolute value sum (FP) of forward prediction;
The difference absolute value sum (BP) of the Backward prediction is calculated, the difference absolute value sum (IPP) of the Inter Position prediction is calculated, and the difference absolute value sum (FP) of the forward prediction and the difference absolute value sum ( BP) and Int
sum of absolute difference (IPP) of er Position prediction
If the smallest sum of absolute differences is larger than the threshold, the prediction mode (Intele) corresponding to the sum of absolute differences (IPP) of the Inter Position prediction is calculated.
r Position prediction) from the prediction mode selection.