JP2012156575A - Encoder, decoder and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To adaptively limit a motion vector search range of a decoding side.SOLUTION: The encoder includes: a motion vector detecting section 15 searching for a reference image and detecting a motion vector; an image compression section 11 compressing a difference image obtained by subtracting a predicted image based on the motion vector from an original image by using prescribed quantization step width; and a quantization step width deciding section 17 updating the quantization step width in the reference region of the motion vector. The quantization step width deciding section 17 updates the quantization step width of the reference region of the motion vector until a prescribed condition is satisfied. The image compression section 11 compresses the original image and compresses the reference image by using the updated quantization step width whenever the quantization step width is updated by the quantization step width deciding section 17. The motion vector detecting section 15 detects the motion vector by referring to the image obtained by decoding the compressed reference image whenever the reference image is compressed by the image compression section 11.

Description

  The present invention relates to an encoding device, a decoding device, and a program that adaptively limit a motion vector search range.

  Motion vector detection is one of the elemental techniques used in video encoding and high frame rate processing by motion compensation interpolation, and blocks similar to the encoding target block (highly correlated) are encoded. This is a technique for detecting by referring to a frame different from the frame to which the block belongs (encoding target frame). In fractal encoding or the like, a block similar to the encoding target block is detected with reference to the same or different frame as the encoding target frame. Generally, detecting a block similar to the encoding target block with reference to a frame different from the encoding target frame is referred to as motion vector detection, and when detecting with reference to the encoding target frame, a template The terms matching and pattern matching are often used. In this specification, the term motion vector detection is used regardless of whether the frame to be referenced is the same as or different from the encoding target frame.

  Also, since motion vector detection is a process with a high calculation cost, it is desirable to reduce the process by limiting the motion vector search range as much as possible. However, if the motion vector search range is uniformly fixed, a correct motion vector may not be included in the search range. Therefore, for example, there is known a motion vector detection device that determines the size of a motion vector detected in the past, widens a motion vector search range for a large motion, and narrows a motion vector search range for a small motion (Patent Literature). 1). In addition, in motion vector detection between frames, a motion vector detection device is known that reduces the motion vector search range as the correlation at the same position increases (see Patent Document 2). Also, there is known a video encoding device that determines a motion vector search range of the next frame based on a total value of absolute values of motion vectors of the previous frame and a total value of prediction errors of the previous frame (Patent Literature). 3).

Japanese Patent Laid-Open No. 10-4554 JP-A-8-32969 JP 2005-244542 A

  However, in the conventional motion vector detection method that adaptively limits the motion vector search range described in Patent Documents 1 to 3, the motion vector is referred to the encoding target frame in order to use temporal correlation of motion vectors. There is a problem that it cannot be applied to a coding system that performs detection.

  Further, in the technique described in Patent Document 1, there is a problem in that a correct motion vector search range cannot be set when erroneous detection is performed on a past motion vector or when the moving speed of an object changes. In the technique described in Patent Document 2, Since the magnitude of the correlation at the same position is affected by noise and object boundaries, there is a problem that a correct motion vector search range may not be obtained.

  In order to solve the above-described problem, an object of the present invention is applicable to an encoding method that performs motion vector detection with reference to an encoding target frame, and adaptively limits a motion vector search range. Another object is to provide a decoding device and a program.

  In image coding and video coding, the quantization step width can be changed for each block. Further, the motion vector does not indicate all the regions in the frame. In order to solve the above problem using this, the encoding apparatus changes the quantization step width of the region referred to by the motion vector. In the decoding apparatus, only the region in which the quantization step width is changed is limited to the motion vector search range.

  That is, in order to solve the above problem, an encoding apparatus according to the present invention is an encoding apparatus that compresses an image and transmits an encoded stream including a quantization step width and quantized image data. A motion vector detection unit that searches for an image to detect a motion vector, an image compression unit that compresses a difference image obtained by subtracting a predicted image based on the motion vector from an original image using a predetermined quantization step width, and A quantization step width determination unit that updates a quantization step width in a motion vector reference region, wherein the quantization step width determination unit quantizes the motion vector reference region until a predetermined condition is satisfied. The step width is updated, and the image compression unit updates the quantization step width every time the quantization step width determination unit updates the quantization step width after compressing the original image. The reference vector is compressed using a quantization step width, and the motion vector detection unit refers to an image obtained by decoding the compressed reference image every time the reference image is compressed by the image compression unit. To detect a motion vector.

  Furthermore, in the encoding device according to the present invention, the predetermined condition is a condition that the number of detections of the motion vector is a predetermined number or more, or the reference region of the motion vector detected repeatedly is detected immediately before. It is a condition that the motion vector is included in the reference region of the motion vector.

  Furthermore, in the encoding device according to the present invention, the quantization step width determination unit updates the quantization step width of the reference region of the updated motion vector to a value smaller than an original value. .

  In order to solve the above problem, a decoding apparatus according to the present invention is a decoding apparatus that receives an encoded stream including a quantization step width and quantized image data, and decodes an image, wherein the encoding An image decoding unit that obtains a quantization step width from the stream and uses the quantization step width to decode an image from the quantized image data, and determines a size of the quantization step width, and the determination A quantization step width determination unit that determines a motion vector search range based on a result; and a motion vector detection unit that detects a motion vector by searching within the motion vector search range in a reference image. Only the changed area is set as a motion vector search range.

  Further, in the decoding device according to the present invention, the quantization step width determination unit determines whether or not the quantization step width is smaller than a predetermined threshold, and moves the region in which the quantization step width is smaller than the predetermined threshold. The vector search range is determined.

  Furthermore, in the decoding device according to the present invention, the quantization step width determination unit is configured to sequentially perform the motion from the region with the smallest quantization step width until a ratio of tightening to the entire frame of the motion vector search range exceeds a predetermined threshold. The vector search range is determined.

  The present invention also provides a computer functioning as an encoding device that compresses an image and transmits an encoded stream including a quantization step width and quantized image data. (B) compressing a difference image obtained by subtracting the predicted image based on the motion vector from the original image using a predetermined quantization step width, and (c) until a predetermined condition is satisfied. Updating the quantization step width in the reference region of the motion vector; and (d) after the step (a), each time the quantization step width is updated by the step (c) Compressing the reference image using a quantization step width; and (e) each time the reference image is compressed in step (d), the compressed reference image Detecting a motion vector by referring to the decoded image, also characterized as a program to be run.

  In addition, the present invention provides a computer functioning as a decoding device that receives an encoded stream including a quantization step width and quantized image data and decodes an image, and (a) a quantization step width from the encoded stream. And decoding the image from the quantized image data using the quantization step width, and (b) determining the size of the quantization step width, and moving based on the determination result It is also characterized as a program for executing a step of determining a vector search range and a step of (c) searching for a motion vector in the motion vector search range in a reference image.

  According to the present invention, the encoding device can instruct the motion vector search range to the decoding device by changing the quantization step width of the region referred to by the motion vector. Since the quantization step width is set for each block of all frames, an appropriate motion vector search range can be adaptively designated for each frame.

  According to the present invention, since the decoding device only needs to set the motion vector search range only for the region where the quantization step width is changed, the calculation cost of motion vector detection can be reduced.

  According to the present invention, since only the quantization step width is used to indicate the motion vector search range, the present invention can also be applied to an encoding method that detects a motion vector with reference to an encoding target frame.

It is a block diagram which shows the structure of the encoding apparatus of one Example by this invention. It is a flowchart which shows operation | movement of the encoding apparatus of one Example by this invention. It is a figure explaining the quantization step width of the encoding apparatus of one Example by this invention. It is a figure explaining operation | movement of the encoding apparatus of one Example by this invention. It is a block diagram which shows the structure of the decoding apparatus of one Example by this invention. It is a flowchart which shows operation | movement of the decoding apparatus of one Example by this invention.

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

[Encoder]
FIG. 1 is a block diagram showing a configuration of an encoding apparatus according to an embodiment of the present invention. The encoding device 1 includes a subtraction unit 10, an image compression unit 11, an image decoding unit 12, an addition unit 13, a decoded image storage unit 14, a motion vector detection unit 15, a motion compensation unit 16, and a quantization. A step width determination unit 17, a search range storage unit 18, and a variable length encoding unit 19 are provided.

  The subtraction unit 10 generates a difference image between the input original image and the predicted image input from the motion compensation unit 16 and outputs the difference image to the image compression unit 11.

  The image compression unit 11 performs orthogonal transform (for example, DCT; Discrete Cosine Transform) processing for each block of a predetermined size (for example, 16 × 16 pixels or 8 × 8 pixels) on the difference image input from the subtraction unit 10. To generate orthogonal transform coefficients. Then, the orthogonal transform coefficient is quantized using the quantization step width input from the quantization step width determination unit 17 to generate differential image compression information (quantized orthogonal transform coefficient). And output to the image decoding unit 12 and the variable length encoding unit 19.

  The image decoding unit 12 performs an inverse quantization process on the quantized orthogonal transform coefficient input from the image compression unit 11. Then, inverse orthogonal transform (for example, IDCT; Inverse Discrete Cosine Transform) is performed on the inversely quantized orthogonal transform coefficient, and a decoded image of the difference image is generated and output to the adding unit 13.

  The adding unit 13 adds the decoded image of the difference image input from the image decoding unit 12 and the predicted image input from the motion compensation unit 16 to generate a decoded image of the original image, and stores the decoded image in the decoded image storage unit 14. Output.

  The decoded image storage unit 14 stores the decoded image input from the addition unit 13.

  The motion vector detection unit 15 searches the motion vector search range stored in the search range storage unit 18 using the reference image (decoded image) stored in the decoded image storage unit 14 as a motion vector detection target, The motion vector of the conversion target block is detected.

  The motion compensation unit 16 performs motion compensation on the reference image input from the decoded image storage unit 14 using the motion vector input from the motion vector detection unit 15, and subtracts the predicted image obtained as a result. To the unit 10 and the adder 13.

  The quantization step width determination unit 17 updates the quantization step width based on the motion vector input from the motion vector detection unit 15 and outputs the updated quantization step width to the image compression unit 11. Further, the quantization step width determination unit 17 may also update the motion vector search range based on the motion vector input from the motion vector detection unit 15. In this case, the updated motion vector search range is output to the search range storage unit 18. By updating the motion vector search range and making the motion vector search range narrower than the entire frame, the processing speed of the motion vector detection unit 15 can be improved.

  The search range storage unit 18 stores the motion vector search range input from the quantization step width determination unit 17. If the quantization step width determination unit 17 does not update the motion vector search range and, for example, the entire frame is always set as the motion vector search range, the search range storage unit 18 need not be provided.

  The variable length encoding unit 19 uses the variable length for the compression information (quantized orthogonal transform coefficient) of the difference image input from the image compression unit 11 and the quantization step width updated by the quantization step width determination unit 17. An encoding process is performed to generate an encoded stream, which is output to the outside. Note that the encoding apparatus 1 of the present embodiment assumes that the decoding apparatus side also detects a motion vector, as will be described later, so there is no need to transmit the motion vector detected by the motion vector detection unit 15. Thereby, the transmission amount can be reduced and the encoding efficiency can be improved.

  Next, the operation of the encoding apparatus 1 configured as described above will be described with reference to FIG. FIG. 2 is a flowchart showing the operation of the encoding apparatus 1.

  In step S101, the quantization step width determination unit 17 determines the quantization step width and the motion vector search range. The initial value of the quantization step width is a uniform value within the frame, and the initial value of the motion vector search range is the entire frame. As described above, the motion vector search range may always be the entire frame.

  In step S102, the motion vector detection unit 15 reads the reference image from the decoded image storage unit 14, sets the motion vector search range stored in the search range storage unit 18 for the reference image, and detects the motion vector. To do.

  In step S103, the motion compensation unit 16 performs motion compensation using the motion vector detected in step S105, and generates a predicted image.

  In step S <b> 104, the subtraction unit 10 generates a difference image between the input original image and the predicted image input from the motion compensation unit 16.

  In step S105, the image compression unit 11 performs orthogonal transform processing and quantization processing on the difference image generated in step S101, and generates compression information (quantized orthogonal transform coefficient) of the difference image. .

  In step S106, the image decoding unit 12 performs inverse quantization processing and inverse orthogonal transformation processing on the quantized orthogonal transform coefficient generated in step S102 to generate a decoded image of the difference image.

  In step S107, the addition unit 13 adds the predicted image generated in step S103 and the decoded image of the difference image generated in step S106 to generate a decoded image of the original image.

  In step S108, the quantization step width determination unit 17 or the control unit (not shown) determines whether or not the number of iterations has reached a predetermined number, or a region (hereinafter referred to as a motion vector detection). It is determined whether or not the motion vector reference region has converged. If the determination result is Yes, the process proceeds to step S109. If the determination result is No, the process from step S101 to step S107 is repeated.

  When iterative processing is performed, the quantization step width determination unit 17 sets the motion vector reference area as a new motion vector search range and outputs the new motion vector search range to the search range storage unit 18. Further, the quantization step width determination unit 17 updates the quantization step width value of each block so that the motion vector reference region can be identified from the region not referenced by the motion vector on the decoding side. Then, the encoding device 1 repeatedly performs the process of compressing the reference image using the quantization step width updated by the quantization step width determination unit 17. The quantization step width determination unit 17 updates the quantization step width until a predetermined condition is satisfied. Here, the predetermined condition is, for example, a condition that the number of iterations is greater than or equal to a predetermined number, or the reference region of the motion vector detected repeatedly is the reference region of the motion vector detected immediately before. It is included (the motion vector reference area converges).

  In step S109, the variable length encoding unit 19 encodes the compression information generated in step S105 to generate an encoded stream.

  FIG. 3 is a diagram for explaining the quantization step width of the encoding device 1. The encoding device 1 performs processing in units of blocks having a predetermined size (for example, 16 × 16 pixels or 8 × 8 pixels). FIG. 3 shows a frame divided into a plurality of blocks. As shown in FIG. 3A, the quantization step width determination unit 17 determines the same quantization step width for all the blocks in the region A of one frame as an initial value. In FIG. 3A, the initial value of the quantization step width is 50. Depending on the encoding method, the size of the block that performs the quantization process may differ from the size of the block that performs the motion compensation process. However, the block that changes the quantization step width in the present invention is a block that performs the quantization process. Refers to that.

  Note that the value determined by the quantization step width determination unit 17 is not limited to the quantization step width itself, and may be information that can express the quantization step width. For example, a plurality of types of quantization step widths may be defined in advance, and the information indicating which type of quantization step width is among them may be used.

  The encoding device 1 performs a first compression process on an input original image of one frame using an initial quantization step width. The motion vector detection unit 15 performs the first motion vector detection, and the quantization step width determination unit 17 determines the motion vector reference region B referred to when the first motion vector is detected. Then, the quantization step width determination unit 17 updates the quantization step width of each block so that on the decoding side, the motion vector reference region can be distinguished from the region that is not referenced by the motion vector. In FIG. 3B, the motion vector reference area B is indicated by a thick frame, and the quantization step width of each block in the motion vector reference area B is changed from 50 to 40.

  When motion vector detection is performed with reference to a decoded image, the accuracy of motion vector detection is improved when the decoding degradation of the decoded image is small. Coding degradation decreases as the quantization step width decreases. Therefore, the quantization step width determination unit 17 preferably changes the quantization step width in the new motion vector search range to a value smaller than the original value.

  The encoding device 1 performs a second compression process on the same reference image of one frame using the updated quantization step width shown in FIG. The motion vector detection unit 15 performs the second motion vector detection, and the quantization step width determination unit 17 determines the motion vector reference region C referred to at the time of the second motion vector detection. In FIGS. 3C and 3D, the motion vector reference region C is indicated by hatching. In the example of FIG. 3C, since the motion vector reference area C is an area included in the motion vector reference area B, it is assumed that the motion vector reference area has converged, and the quantization step width determination unit 17 The update of the quantization step width of the reference area is finished.

  On the other hand, in the example of FIG. 3D, since the motion vector reference area C is an area not included in the motion vector reference area B, the quantization step width determination unit 17 determines the quantization step width of the motion vector reference area. Update. FIG. 3D shows a state in which the quantization step width of each block in the motion vector reference region C is changed to 30. Note that if the quantization step width is greatly changed from the initial value, the influence on the bit rate is increased. Therefore, as shown in FIG. 3E, only the quantization step width in the newly updated motion vector reference region C may be set to 40, and the others may be set to 50, which is the same as the initial value.

  In FIGS. 3B to 3E, the motion vector reference regions B and C are shown as a single region, but the motion vector reference region may be a plurality of regions or intermittent. It may be an area where a portion exists. Alternatively, the motion vector detection unit 15 may initially search roughly within the motion vector search range and finely search as the motion vector search range becomes narrower.

  The encoding device 1 can perform motion compensation with reference to the encoding target frame, or can perform motion compensation with reference to a frame different from the encoding target frame. FIG. 4A is a diagram for explaining the operation of the encoding apparatus 1 when motion compensation is performed with reference to the encoding target frame, and FIG. 4B refers to a frame different from the encoding target frame. It is a figure explaining operation | movement of the encoding apparatus 1 in the case of performing motion compensation.

  As shown in FIG. 4A, when motion compensation is performed with reference to the encoding target frame 1, first, with respect to the encoding target frame 1, the detection unit 15 searches the entire encoding target frame 1 for a motion vector. The motion vector a is detected as a range, and the image compression unit 11 compresses the encoding target frame 1 using the same value (50 in the example of FIG. 3) in the entire encoding target frame 1. Next, the quantization step of the motion vector reference region B referred to when the motion vector a is detected is changed by the image compression unit 11 (40 in the example of FIG. 3), and the encoding target frame 1 which is a reference image is changed. And compress again.

  Here, since the quantization step width of the motion vector reference region B is changed and compressed, the decoded image by the image decoding unit 12 changes compared to the original decoded image. Therefore, in order to confirm whether the motion vector reference region in this case is a region included in the motion vector reference region B, the motion vector detection unit 15 sets the entire encoding target frame 1 as a motion vector search range, The motion vector b is detected for the decoded image by the image decoding unit 12. However, it is not necessary to re-detect the motion vector for a block that refers to an area where the decoded image has not changed.

  When the motion vector reference region C referred to when the motion vector b is detected is included in the motion vector reference region B, the compression encoding process for the encoding target frame 1 is terminated. On the other hand, when the motion vector reference region C is not included in the motion vector reference region B, the quantization step of the motion vector reference region C is changed, the compression is performed again on the encoding target frame 1, and then Similarly, the motion vector detection unit 15 detects the motion vector c for the decoded image by the image decoding unit 12 using the entire encoding target frame 1 as a motion vector search range. In this way, this process is repeated until the number of iterations reaches a predetermined number of times or the motion vector search range converges.

  As shown in FIG. 4B, when motion compensation is performed with reference to a decoded frame different from the encoding target frame 1, first, the entire decoded frame 2 is detected by the detection unit 15 for the encoding target frame 1. The motion vector a is detected as a motion vector search range, and the image compression unit 11 compresses the encoding target frame 1 using the same value (50 in the example of FIG. 3) in the entire encoding target frame 1. Next, the quantization step of the motion vector reference region B referred to when the motion vector a is detected is changed by the image compression unit 11 (40 in the example of FIG. 3), and the decoded frame 2 that is the reference image is changed. Perform compression.

  Here, since the quantization step width of the motion vector reference region B is changed and compressed, the decoded image by the image decoding unit 12 changes compared to the original decoded image. Therefore, in order to confirm whether the motion vector reference area in this case is an area included in the motion vector reference area B, the motion vector detection unit 15 sets the entire decoded frame 2 as a motion vector search range, The motion vector b is detected from the decoded image by the decoding unit 12. However, it is not necessary to re-detect the motion vector for a block that refers to an area where the decoded image has not changed.

  When the motion vector reference region C referred to when the motion vector b is detected is included in the motion vector reference region B, the compression encoding process for the decoded frame 2 is ended. On the other hand, when the motion vector reference region C is not included in the motion vector reference region B, the quantization step of the motion vector reference region C is changed, the decoded frame 2 is compressed again, and then the same The motion vector detection unit 15 detects the motion vector c for the decoded image by the image decoding unit 12 using the entire decoded frame 2 as the motion vector search range. In this way, this process is repeated until the number of iterations reaches a predetermined number of times or the motion vector search range converges.

  In this way, the encoding device 1 performs the compression process twice or more and updates the quantization step width, thereby limiting the motion vector search range on the decoding side. Although the possibility is very low, if the reference destination of the motion vector fills the entire area of the frame of the reference image, an appropriate motion vector cannot be found unless the entire frame is searched. The vector search range cannot be limited. Therefore, in such a case, the quantization step width is exceptionally constant in the entire region in the frame.

  In addition, the present invention can be easily combined with other motion vector search range limiting methods. For example, the above-described iterative process is performed using a motion vector search range set by another method as an initial value of the motion vector search range, or another motion vector search range limiting method and a motion vector reference region according to the present invention are obtained, It is also possible to determine a region included in both as a final motion vector search range on the decoding side.

  In general, the reference destination of the motion vector does not indicate all the areas in the frame, and there is an unreferenced area. Therefore, the encoding device 1 can adaptively limit the motion vector search range on the decoding side within the motion vector reference region of the encoding device 1.

[Decryption device]
Next, a decoding apparatus according to an embodiment of the present invention will be described. FIG. 5 is a block diagram showing a configuration of a decoding apparatus according to an embodiment of the present invention. The decoding device 2 includes a variable length decoding unit 21, an image decoding unit 22, an addition unit 23, a decoded image storage unit 24, a motion vector detection unit 25, a motion compensation unit 26, and a quantization step width determination unit 27. And a search range storage unit 28.

  The variable length decoding unit 21 receives the encoded stream from the encoding device 1. The encoded stream input from the encoding device 1 includes compression information (quantized orthogonal transform coefficient) of the difference image and a quantization step width. The variable length decoding unit 21 performs variable length decoding processing on the input encoded bitstream, and acquires compression information of the difference image and a quantization step width. Then, the compression information of the difference image is output to the image decoding unit 22, and the quantization step width is output to the image decoding unit 22 and the quantization step width determination unit 27.

  The image decoding unit 22 uses the quantization step width input from the variable length decoding unit 21 to perform inverse quantization processing on the compression information of the difference image input from the variable length decoding unit 21. Then, inverse orthogonal transform processing is performed on the inversely quantized orthogonal transform coefficient, a decoded image of the difference image is generated, and output to the adding unit 23.

  The addition unit 23 adds the decoded image of the difference image input from the image decoding unit 22 and the predicted image input from the motion compensation unit 26 to generate a decoded image of the original image, and outputs the decoded image storage unit 24 and Output to the outside.

  The decoded image storage unit 24 stores the decoded image input from the adding unit 23.

  The motion vector detection unit 25 detects a motion vector by searching the motion vector search range stored in the search range storage unit 28 using the reference image (decoded image) stored in the decoded image storage unit 24 as a motion vector detection target. To do. There are many motion vector detection methods, such as a block matching method, gradient method, phase correlation method, and speeding-up methods thereof, but these are used on the encoding side in order to obtain the same motion vector as the encoding side. The same detection method as the motion vector detection method is used.

  The motion compensation unit 26 performs motion compensation on the reference image input from the decoded image storage unit 24 using the motion vector input from the motion vector detection unit 25, and adds the predicted image obtained as a result to the addition unit. To 23.

  The quantization step width determination unit 27 determines the size of the quantization step width input from the variable length decoding unit 21, and determines a motion vector search range from the quantization step width.

  The search range storage unit 28 stores the motion vector search range input from the quantization step width determination unit 27.

  Next, the operation of the decoding apparatus 2 configured as described above will be described with reference to FIG. FIG. 6 is a flowchart showing the operation of the decoding device 2.

  In step S201, the variable length decoding unit 21 performs a variable length decoding process on the input encoded bitstream to obtain a quantization step width.

  In step S202, the variable-length decoding unit 21 performs variable-length decoding processing on the input encoded bitstream, and acquires compression information of the difference image.

  In step S <b> 203, the image decoding unit 22 decompresses the compressed image of the difference image acquired by the variable length decoding unit 21 to generate a decoded image.

  In step S204, the quantization step width determination unit 27 determines the quantization step width acquired in step S201, and determines a region in which the quantization step width is changed as a motion vector search range. As described above, it is preferable that the encoding apparatus 1 changes the quantization step width in the new motion vector search range to a value smaller than the original value, and the decoding apparatus 2 performs encoding in such a preferable form. When an encoded stream is received from the apparatus 1, an area having a small quantization step width is determined as a motion vector search range.

  The size of the quantization step width is determined by whether or not a predetermined threshold is exceeded, or by whether or not the ratio of the frame being decoded to the minimum quantization step width exceeds a predetermined threshold. Can be determined by various methods, such as determining whether the ratio of fastening to the entire frame of the motion vector search range exceeds a predetermined threshold when the motion vector search range is set in ascending order of quantization step width. It is. Further, the determination method and the threshold used for the determination may be determined in advance, or may be transmitted for each frame or each sequence.

  In step S205, the motion vector detection unit 25 reads the reference image from the decoded image storage unit 24, sets the motion vector search range stored in the search range storage unit 28 for the reference image, and detects the motion vector. To do.

  In step S206, the motion compensation unit 26 performs motion compensation using the motion vector detected in step S204, and generates a predicted image.

  In step S207, the addition unit 23 adds the decoded image of the difference image generated in step S205 and the predicted image input from the motion compensation unit 26 to generate a decoded image.

  When motion vector detection is performed with reference to a decoded image, the accuracy of motion vector detection is improved when the decoding degradation of the decoded image is small. Coding degradation decreases as the quantization step width decreases. Since the decoding apparatus 2 uses only a region having a small quantization step width as a motion vector search range, it is possible to improve motion vector detection accuracy. Furthermore, according to the present invention, since the region referred to by the motion vector has a small quantization step width and little coding deterioration, the image quality of the decoded image can be improved.

  Here, in order to function as the encoding device 1, a computer can be suitably used. By causing the computer to execute a predetermined program, the subtraction unit 10, the image compression unit 11, and the image decoding unit 12, an addition unit 13, a decoded image storage unit 14, a motion vector detection unit 15, a motion compensation unit 16, a quantization step width determination unit 17, a search range storage unit 18, and a variable length encoding unit 19 It is possible to realize the functions of

  Similarly, a computer can be suitably used for causing the computer to function as the decoding device 2, and by causing the computer to execute a predetermined program by the CPU, the variable length decoding unit 21, the image decoding unit 22, and the addition unit 23, the decoded image storage unit 24, the motion vector detection unit 25, the motion compensation unit 26, the quantization step width determination unit 27, and the search range storage unit 28 can be realized.

  Although the above embodiments have been described as representative examples, it will be apparent to those skilled in the art that many changes and substitutions can be made within the spirit and scope of the invention. Therefore, the present invention should not be construed as being limited by the above-described embodiments, and various modifications and changes can be made without departing from the scope of the claims.

  As described above, according to the present invention, the motion vector search range can be adaptively limited, which is useful for any application for efficiently encoding and decoding images and videos.

DESCRIPTION OF SYMBOLS 1 Encoding apparatus 2 Decoding apparatus 10 Subtraction part 11 Image compression part 12 Image decoding part 13 Addition part 14 Decoded image memory | storage part 15 Motion vector detection part 16 Motion compensation part 17 Quantization step width determination part 18 Search range memory | storage part 19 Variable length Encoding unit 21 Variable length decoding unit 22 Image decoding unit 23 Addition unit 24 Decoded image storage unit 25 Motion vector detection unit 26 Motion compensation unit 27 Quantization step width determination unit 28 Search range storage unit

Claims (8)

An encoding device for compressing an image and transmitting an encoded stream including a quantization step width and quantized image data,
A motion vector detection unit that searches for a reference image and detects a motion vector;
An image compression unit that compresses a difference image obtained by subtracting a predicted image based on the motion vector from an original image using a predetermined quantization step width;
A quantization step width determination unit that updates a quantization step width in a reference region of the motion vector, and
The quantization step width determination unit updates the quantization step width of the reference region of the motion vector until a predetermined condition is satisfied,
The image compression unit compresses the reference image using the updated quantization step width every time the quantization step width is updated by the quantization step width determination unit after compressing the original image. ,
The motion vector detection unit detects a motion vector with reference to an image obtained by decoding the compressed reference image every time the reference image is compressed by the image compression unit.   The predetermined condition is that the number of detections of the motion vector is a predetermined number or more, or the reference region of the motion vector detected repeatedly is included in the reference region of the motion vector detected immediately before. The encoding apparatus according to claim 1, wherein the condition is a condition.   The encoding apparatus according to claim 1, wherein the quantization step width determination unit updates a quantization step width of a reference region of the motion vector to a value smaller than an original value. A decoding device for receiving an encoded stream including a quantization step width and quantized image data and decoding an image,
An image decoding unit that obtains a quantization step width from the encoded stream and uses the quantization step width to decode an image from the quantized image data;
A quantization step width determination unit that determines the size of the quantization step width and determines a motion vector search range based on the determination result;
A motion vector detection unit for detecting a motion vector by searching within the motion vector search range in a reference image;
A decoding apparatus comprising:   The quantization step width determination unit determines whether or not the quantization step width is smaller than a predetermined threshold, and determines an area where the quantization step width is smaller than the predetermined threshold as the motion vector search range. The decoding device according to claim 4.   The quantization step width determination unit determines the motion vector search range in order from the region with the smallest quantization step width until the ratio of fastening to the entire frame of the motion vector search range exceeds a predetermined threshold. The decoding device according to claim 4. To a computer functioning as an encoding device that compresses an image and transmits an encoded stream including a quantization step width and quantized image data,
(A) searching for a reference image to detect a motion vector;
(B) compressing a difference image obtained by subtracting a predicted image based on the motion vector from an original image using a predetermined quantization step width;
(C) updating a quantization step width in a reference region of the motion vector until a predetermined condition is satisfied;
(D) After the step (a), each time the quantization step width is updated by the step (c), the step of compressing the reference image using the updated quantization step width;
(E) detecting a motion vector with reference to an image obtained by decoding the compressed reference image every time the reference image is compressed in the step (d);
A program for running A computer functioning as a decoding device that receives an encoded stream including a quantization step width and quantized image data and decodes the image;
(A) obtaining a quantization step width from the encoded stream, and decoding the image from the quantized image data using the quantization step width;
(B) determining a size of the quantization step width, and determining a motion vector search range based on the determination result;
(C) searching a motion vector search range in a reference image to detect a motion vector;
A program for running

Images