JP2011223393A - Encoding device and method of controlling encoding device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce deterioration in picture quality by improving detection precision of a motion vector for a repetitive pattern.SOLUTION: An encoding device comprises global vector acquiring means of acquiring a motion vector for each of a plurality of second blocks constituting a first block, prediction error value calculating means of calculating a vector value at an address where encoding is performed, vector code amount calculating means of calculating a vector code amount for the address where the encoding is performed, vector evaluated value calculating means of calculating a vector evaluated value by weighting the global vector code amount calculated by the vector code amount calculating means, and vector position determining means of determining a vector position based upon the vector evaluated value calculated by the vector evaluated value calculating means, thereby eliminating trouble such that a vector different from actual motion is selected even for a repetitive pattern such as a window frame.

Description

  The present invention relates to an encoding apparatus and an encoding apparatus control method, and more particularly to an encoding apparatus for determining a motion vector.

  In an encoding apparatus that encodes a moving image, there are various methods for encoding an image (repeated pattern image) in which similar patterns (small images) periodically exist, such as windows of a building or a lattice frame. It is being considered. In particular, development of an encoding apparatus having a countermeasure function against vector misdetection that occurs when there is a motion in a repetitive pattern image has been studied.

  In vector search, there is a method of determining a motion vector from a minimum SAD (Sum of Absolute Difference) between an encoding target block and a reference block. However, in the repeated pattern image, the position of the minimum SAD is not necessarily the motion vector position because of the same pattern, and thus vector misdetection occurs.

  Regarding countermeasures against erroneous vector detection of repetitive pattern images, the interpolation frame creation apparatus described in Patent Document 1 compares a SAD minimum point obtained with a variable-length block when there are a plurality of SAD minimum points of a block. A vector corresponding to the minimum point closest to the SAD value obtained using the variable length block is employed.

JP 2008-147951 A

  However, in the configuration of the interpolated frame creation device proposed in Patent Document 1, even if the nearest SAD is the closest SAD using the SAD minimum point of the variable length block larger than the unit block, the motion is not always correct. There is a problem that cannot be detected.

In particular, in the case of image movement when a repeated pattern image of the same pattern such as a window frame of a building is taken by panning or tilting, a motion vector indicating the wrong place is detected even if the SAD is minimum, thereby degrading image quality. There is a possibility of letting you.
The present invention has been made in view of the above-described problems, and an object of the present invention is to improve the accuracy of motion vector detection for repeated pattern images and reduce image quality degradation.

An encoding apparatus according to the present invention is an encoding apparatus that divides an input image signal into first blocks of a predetermined size and performs encoding in units of the first blocks, and includes a plurality of the first blocks. Global vector acquisition means for acquiring a motion vector for each second block to be performed; prediction error value calculation means for calculating a prediction error value at an address to be encoded;
Vector code amount calculation means for calculating a vector code amount at the address to be encoded, and vector evaluation value calculation means for calculating a vector evaluation value by weighting the global vector code amount calculated by the vector code amount calculation means And vector position determining means for determining a vector position based on the vector evaluation value calculated by the vector evaluation value calculating means.

  According to the present invention, it is possible to reduce erroneous detection of motion vectors caused by the motion of repetitive pattern images, and to suppress image quality deterioration.

It is a block diagram which shows 1st Embodiment and shows the structural example of an encoding apparatus. It is a figure which shows the example which a building moves in the image by panning. It is a flowchart explaining the structural example and operation | movement of a motion search part. It is a flowchart explaining the process inside a motion search part. It is a figure which shows the concept of the weighting at the time of pan / tilt effective. It is a figure showing the pixel block at the time of SAD calculation. It is a flowchart which shows 2nd Embodiment and shows the process of a motion search part. It is a figure explaining the structure from the structure of a motion search part, and the distance from the minimum value position of SAD.

Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings.
(First embodiment)
A first embodiment will be described with reference to FIGS.
FIG. 1 is a block diagram illustrating a configuration example of the encoding apparatus according to the present embodiment. The encoding apparatus according to the present embodiment divides an input image signal into first blocks having a predetermined size and performs encoding in units of first blocks.

  In FIG. 1, a line input terminal 101 inputs a video signal output from an image sensor provided in a digital video camera, for example. The encoding unit 102 performs encoding corresponding to standards such as MPEG-2 and MPEG-4 AVC, for example. The decoding unit 103 decodes the stream encoded by the encoding unit 102. Reference numeral 104 denotes a bus. Reference numeral 105 denotes an information input terminal for inputting image information input from the outside or information output from a pan / tilt detector (not shown). An image memory 106 stores an input image and a local decoded image decoded by the decoding unit 103. A motion search unit 107 determines a motion vector used in the encoding unit 102. An output terminal 108 outputs the stream encoded by the encoding unit 102.

FIG. 2A is an example of an encoded reference image, and FIG. 2B is a diagram illustrating an example of an encoding target image. For example, two images that change from FIG. 2A to FIG. 2B can be obtained by taking a picture while the digital video camera pans to the left. At this time, the building is moving in the image by panning. FIG. 2 (c) is a diagram showing candidates for the motion position of a certain macroblock.
As shown in FIG. 2, in a repetitive pattern in which the same pattern continues, when matching a reference image by SAD (Sum of Absolute Difference), a plurality of local minimum values are generated within the search range as shown in FIG. . The minimum value of SAD may select the wrong place and cause image quality degradation.

FIG. 3 is a block diagram illustrating an example of an internal configuration of the motion search unit 107.
As shown in FIG. 3, the motion search unit 107 of this embodiment includes a pan / tilt emphasis determination unit 301, an SAD calculation unit 302, an evaluation value calculation unit 303, and a vector determination unit 304. The pan / tilt emphasis determining unit 301 performs a process once on an area block (second block) composed of one picture or a plurality of macroblocks (first block). That is, global vector acquisition processing is performed for each second block, and processing is performed for each first block otherwise.

The operation of the pan / tilt emphasis determination unit 301 will be described below.
FIG. 3B is a flowchart for explaining a procedure for changing processing in a picture according to image information or output information from a pan / tilt detector (not shown).
In FIG. 3B, pan / tilt information is acquired in step S3011. This information is, for example, information from a gyro sensor in a video camera. In step S3012, it is determined whether pan / tilt is valid. For example, the determination is made based on whether or not the difference value is equal to or less than a preset threshold at the position obtained from the pan / tilt information between the reduced images of the reference image and the encoding target image.

  If the result of determination in S3012 is that pan / tilt is not valid, the process proceeds to S3013, where pan / tilt priority OFF is set to perform normal processing. If the result of determination in S3012 is that pan / tilt is valid, the process proceeds to S3014, and pan / tilt priority is set to ON and priority vector positions are detected in order to perform processing that places importance on the pan / tilt position.

  At this time, the normal processing outputs the weighting factor α = 1, and the pan / tilt position peripheral emphasis outputs an integer of α> 1 as the weighting factor. The pan / tilt emphasis determination unit 301 performs pan / tilt setting in units of pictures or area blocks.

Next, operations of the three blocks of the SAD calculation unit 302, the evaluation value calculation unit 303, and the vector determination unit 304 will be described.
FIG. 4 is a flowchart showing the processing of the three blocks described above.
First, in step S401, the address is initialized (i = 0), and the quantization coefficient Q at the initial address and the pan / tilt setting (presence / absence of pan / tilt priority) output from the pan / tilt determination unit are acquired.

  Next, in S402, the weight β is determined. The weight β is determined using information output from the pan / tilt emphasis determining unit 301. When the pan / tilt priority is given, the weight β is determined using the pan / tilt position information.

  FIG. 5 is a diagram illustrating the value of the weight β. When the pan / tilt position is a gray portion, the weight changes depending on the distance from the position. However, the upper limit value of the weight β is 10 here. When the pan / tilt priority is OFF, the weight β is a constant value. Here, the weight β = 10 is output.

As described above, after the weight β is determined, the process proceeds to S403. In S403, the SAD calculation unit 302 calculates the block SAD at the address i (prediction error value calculation).
FIG. 6 is a diagram illustrating a pixel block at the time of SAD calculation. FIG. 6A shows a block (current block) in a selected state, and FIG. 6B shows a search range. One square indicated by B [0] to B [24] is one pixel. The address is a position that takes a difference from the upper left pixel in FIG. At this time, a difference absolute value is obtained for each pixel, and the sum is used as a difference absolute value sum SAD.

Next, in S404, a vector code amount calculation process for calculating the vector code amount MV at the address i is performed. Then, it progresses to S405. In S405, the global vector code amount calculated in S404 is weighted. This is a process of assigning a weight according to the distance from the global vector position for each first block. After the global vector weighting process, a vector evaluation value calculation process is performed. Specifically, the evaluation value calculation unit 303 calculates the evaluation value BV [i]. Evaluation value BV [i] is
BV [i] = α * f (Q, SAD) + β * MV (1 set)
It is.

  Here, f (Q, SAD) indicates that a quantization code amount evaluation value is obtained from the quantization coefficient Q and the SAD value. This varies with the quantization matrix. However, the weighting factor α = 10. When there is no pan / tilt priority, evaluation is performed at the same ratio, and when there is pan / tilt priority, the pan / tilt position is prioritized.

  In S406, it is determined whether or not the evaluation value BV [i] calculated in S405 is the smallest among the evaluation values calculated so far. As a result of this determination, if it is the minimum, the minimum value MinBV of the evaluation value is updated in S407, and then the process proceeds to S408. If not the minimum, the process proceeds to S408.

  In S408, it is determined whether or not the entire search range has been processed. If the result of this determination is that processing has not been performed, the address is advanced by one in step S409, and then control returns to step S402 to perform the processing described above. If the processing has been completed, the vector determination unit 304 performs a vector position determination process using the address having the minimum value MinBV at that time as the vector position.

(Second Embodiment)
Next, a second embodiment of the present invention will be described with reference to FIGS.
In the present embodiment, a vector search is performed for a case where there are a plurality of minimum points in the vector search range (third block), and the overall configuration of the encoding apparatus is the same as that of the first embodiment. It is.
FIG. 8A is a block diagram illustrating an internal configuration example of the motion search unit 107 in the present embodiment. In FIG. 8A, the configurations of the SAD calculation unit 302, the evaluation value calculation unit 303, and the vector determination unit 304 are the same as those in FIG. Reference numeral 805 denotes a pan / tilt information acquisition unit, and reference numeral 806 denotes a local minimum point detection unit.

The operation of the motion search unit 107 in the present embodiment will be described below.
FIG. 7 is a flowchart for explaining the process performed by the five blocks described in FIG. 8A. When the process is started, first, in step S701, the address is initialized (i = 0) and quantized. The coefficient Q is acquired.
Next, proceeding to S702, a process of calculating a block SAD at address i is performed. This process is the same as S403 in the flowchart of FIG. 4 in the first embodiment.

  In step S <b> 703, the minimum point detection unit 806 performs minimum point detection. If there is only one minimum point in the search range, it is appropriate as a vector position. However, in an image such as a repeated pattern, there are a plurality of minimum points (minimum values). For this reason, the minimum point detection is performed in S703. The minimum point detection (minimum value detection) performed in the present embodiment is detected from the distance from the SAD minimum value position and the SAD value within a predetermined search range.

  FIG. 8B is a diagram illustrating the distance from the minimum value position of the SAD. In the case of a normal image, the SAD gradually increases as the distance from the minimum value position of the SAD increases. For this reason, there is the second smallest SAD in any place where the distance is 1, and the larger the distance, the larger the SAD. In the case of a repetitive pattern, SAD close to the minimum value is generated at a location where the distance is long, and therefore, it can be detected by the distance and the size of SAD. When there is an SAD that is close to the minimum value at a distance, the minimum point is two or more.

In S704, it is determined from the detection result in S703 whether or not there is one minimum point. If the result of this determination is that there are two or more local minimum points, the process proceeds to S705, where the pan / tilt information acquisition unit 805 acquires the pan / tilt position. Thereafter, in S706, a weight value determination process for determining a weight value corresponding to the pan / tilt position acquired in S705 is performed to determine the weight β. Thereafter, the process proceeds to S707. The pan / tilt position acquisition processing performed in S705 and the weight β value determination processing performed in S706 are the same as those in the first embodiment.
On the other hand, if the result of determination in S704 is one minimum point, the process proceeds directly to S707. Since the processing after S707 is the same as the processing after S404 of the first embodiment, a description thereof will be omitted.

  As described above, in this embodiment, the minimum point is detected after the block SAD at the address i is calculated, and in the macro block having two or more minimum points, the movement obtained from the pan / tilt information is emphasized. Thus, the weight β is determined. As a result, it is possible to prevent the erroneous detection of the vector due to the repetitive pattern, thereby eliminating the erroneous detection of the vector and suppressing the image quality deterioration.

  As described above, when there are a plurality of local minimum points in the search range, the position where the evaluation value is the minimum value is used as the motion vector position in the normal vector search process, so a position different from the motion may be selected. . In the present embodiment, as described above, a global vector (GV) that is a global motion vector is detected. Then, the degree of coincidence is determined from the direction, size, and reliability of the global vector GV for each area. A vector search is performed by adding a gradient (weight) to the pixel difference value around the position of the global vector GV, and a motion vector that minimizes the pixel difference value is selected in addition to the position around the position where the prediction error value becomes the minimum value. I tried to do it. Thereby, even if it is a repetitive pattern such as a window frame, it is possible to eliminate the inconvenience of selecting a vector different from the actual motion, and it is possible to suppress deterioration in image quality.

  In the above-described embodiment, the difference absolute value sum SAD is used as the prediction error value. However, a prediction error value such as an absolute value sum SATD (Sum of Absolute Transform Difference) obtained by orthogonal transform of the difference value may be used.

(Other embodiments)
The present invention can also be realized by executing the following processing. That is, software (program) that realizes the functions of the above-described embodiments is supplied to the system or apparatus via a network or various computer-readable storage media. This is a process in which the computer (or CPU, MPU, etc.) of the system or apparatus reads out the program and executes control of each part.

101 line input terminal, 102 encoding unit, 103 decoding unit, 104 bus, 105 information input terminal, 106 image memory, 107 motion search unit, 108 output terminal

Claims (13)

An encoding device that divides an input image signal into first blocks of a predetermined size and performs encoding in units of the first blocks,
Global vector acquisition means for acquiring a motion vector for each second block comprising a plurality of the first blocks;
Prediction error value calculating means for calculating a prediction error value at an address to be encoded;
Vector code amount calculation means for calculating a vector code amount at an address to be encoded;
A vector evaluation value calculating means for calculating a vector evaluation value by weighting the global vector code amount calculated by the vector code amount calculating means;
An encoding apparatus comprising: vector position determining means for determining a vector position based on the vector evaluation value calculated by the vector evaluation value calculating means. An encoding device that divides an input image signal into first blocks of a predetermined size and performs encoding in units of the first blocks,
Global vector acquisition means for acquiring a motion vector for each second block comprising a plurality of the first blocks;
Prediction error value calculating means for calculating a prediction error value at an address to be encoded;
A distance from the minimum position of the prediction error value in the third block, which is a search range for performing vector search, and a minimum value detecting means for detecting a minimum value from the prediction error value;
Determination means for determining whether or not the minimum value detected by the minimum value detection means is one;
Pan / tilt position acquisition means for acquiring a pan / tilt position at the minimum value detected by the minimum value detection means;
A weight value determining means for determining a weight value according to the pan / tilt position acquired by the pan / tilt position acquiring means;
Vector code amount calculation means for calculating a vector code amount at an address to be encoded;
A vector evaluation value calculating means for calculating a vector evaluation value by weighting the global vector code amount calculated by the vector code amount calculating means;
Vector position determining means for determining a vector position based on the vector evaluation value calculated by the vector evaluation value calculating means,
The vector evaluation value calculation means uses a predetermined constant weight value when there is one minimum value as a result of determination by the determination means, and when the determination means determines that there are a plurality of minimum values. The coding apparatus characterized in that a vector evaluation value is calculated by performing weighting using the weight value determined by the weight value determining means. The vector evaluation value is calculated based on the prediction error value and a vector code amount,
The encoding apparatus according to claim 1, wherein the vector position determination unit determines a position where the vector evaluation value is a minimum value as a vector position.   The encoding apparatus according to claim 1, wherein the prediction error value is a sum of absolute differences (SAD).   The encoding apparatus according to claim 1, wherein the prediction error value is an absolute value sum (SATD) obtained by orthogonally transforming a difference value.   The encoding apparatus according to claim 1, wherein the vector evaluation value calculation unit assigns a weight according to a distance from a global vector position for each of the first blocks.   3. The encoding apparatus according to claim 1, wherein when the global vector is 0, the vector evaluation value calculation unit equalizes all of the weights assigned to the first blocks. 4.   3. The encoding according to claim 2, wherein the minimum value detecting unit detects that there is a small prediction error value other than a position around the position where the prediction error value is minimum in the third block. apparatus. A method for controlling an encoding apparatus that divides an input image signal into first blocks of a predetermined size and performs encoding in units of the first blocks,
A global vector acquisition step of acquiring a motion vector for each second block comprising a plurality of the first blocks;
A prediction error value calculating step of calculating a prediction error value at an address to be encoded;
A vector code amount calculating step of calculating a vector code amount at an address to be encoded;
A vector evaluation value calculation step of calculating a vector evaluation value by weighting the global vector code amount calculated in the vector code amount calculation step;
And a vector position determining step of determining a vector position based on the vector evaluation value calculated in the vector evaluation value calculating step. A method for controlling an encoding apparatus that divides an input image signal into first blocks of a predetermined size and performs encoding in units of the first blocks,
A global vector acquisition step of acquiring a motion vector for each second block comprising a plurality of the first blocks;
A prediction error value calculating step of calculating a prediction error value at an address to be encoded;
A distance from the minimum position of the prediction error value in the third block, which is a search range for performing vector search, and a minimum value detection step of detecting a minimum value from the prediction error value;
A determination step of determining whether or not there is one minimum value detected in the minimum value detection step;
A pan / tilt position acquisition step of acquiring a pan / tilt position at the minimum value detected in the minimum value detection step;
A weight value determining step for determining a weight value according to the pan / tilt position acquired in the pan / tilt position acquiring step;
A vector code amount calculating step of calculating a vector code amount at an address to be encoded;
A vector evaluation value calculation step of calculating a vector evaluation value by weighting the global vector code amount calculated in the vector code amount calculation step;
A vector position determining step for determining a vector position based on the vector evaluation value calculated in the vector evaluation value calculating step,
The vector evaluation value calculation step uses a predetermined constant weight value when there is one minimum value as a result of the determination in the determination step, and when the determination step determines that there are a plurality of minimum values. Is a method for controlling a coding apparatus, wherein a vector evaluation value is calculated by performing weighting using the weight value determined in the weight value determination step. A program for causing a computer to execute a process of controlling an encoding apparatus that divides an input image signal into first blocks of a predetermined size and performs encoding in units of the first blocks,
A global vector acquisition step of acquiring a motion vector for each second block comprising a plurality of the first blocks;
A prediction error value calculating step of calculating a prediction error value at an address to be encoded;
A vector code amount calculating step of calculating a vector code amount at an address to be encoded;
A vector evaluation value calculation step of calculating a vector evaluation value by weighting the global vector code amount calculated in the vector code amount calculation step;
A program causing a computer to execute a vector position determination step of determining a vector position based on the vector evaluation value calculated in the vector evaluation value calculation step. A program for causing a computer to execute a process of controlling an encoding apparatus that divides an input image signal into first blocks of a predetermined size and performs encoding in units of the first blocks,
A global vector acquisition step of acquiring a motion vector for each second block comprising a plurality of the first blocks;
A prediction error value calculating step of calculating a prediction error value at an address to be encoded;
A distance from the minimum position of the prediction error value in the third block, which is a search range for performing vector search, and a minimum value detection step of detecting a minimum value from the prediction error value;
A determination step of determining whether or not there is one minimum value detected in the minimum value detection step;
A pan / tilt position acquisition step of acquiring a pan / tilt position at the minimum value detected in the minimum value detection step;
A weight value determining step for determining a weight value according to the pan / tilt position acquired in the pan / tilt position acquiring step;
A vector code amount calculating step of calculating a vector code amount at an address to be encoded;
A vector evaluation value calculation step of calculating a vector evaluation value by weighting the global vector code amount calculated in the vector code amount calculation step;
Causing the computer to execute a vector position determination step for determining a vector position based on the vector evaluation value calculated in the vector evaluation value calculation step;
In the vector evaluation value calculation step, when the result of determination in the determination step is one minimum value, a predetermined weight value is used in advance. When the determination step determines that there are a plurality of minimum values Includes a program for causing a computer to calculate a vector evaluation value by performing weighting using the weight value determined in the weight value determination step.   A computer-readable storage medium storing the program according to claim 11 or 12.

Images