JP2006101397A - Moving image encoding device and moving image encoding program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To obtain a moving image encoding device which enables the detection of the insertion/non-insertion of a loop filter to be easily performed, when a signal encoded by inserting and non-inserting the loop filter is decoded. <P>SOLUTION: An average quantized value of a macro block within a single slice image of a quantized input image is obtained by an average value calculating portion 254 by being compared with a quantized threshold value by a first comparator 256. A dispersed value of the quantized value is obtained by a dispersion value calculating portion 251 by being compared with a dispersion threshold value by a comparator 253. When the comparison result at the first comparator is smaller than a predetermined quantized value and the dispersion value is also smaller than the quantized value, a loop filter is determined to be turned off. When the quantized value and the dispersion value are larger, the loop filter is determined to be turned on by a filter level determining portion 257. When macro blocks where the loop filter immediately before the loop filter determined to be turned on up to the macro block is determined to be turned off sequentially exist, its section is determined to be a new slice section by a slice section determining portion 193, and encoded. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  In the present invention, when an image signal is divided into blocks and subjected to transform coding, a block that is assumed to generate block noise is subjected to block noise removal processing and compression coded in advance, and the compression coded signal is decoded. The present invention relates to a moving image coding apparatus and a moving image coding program for generating a coded signal for generating and decoding a reference image that has been subjected to the same block noise removal processing as that for coding.

  Recently, MPEG (Moving Picture Experts Group) -4 and H.264 have been used as high-efficiency compression encoding methods for images in the fields of communication and storage. H.264 is widely used. In these encoding methods, an encoded signal obtained by removing redundancy related to space and time from an image is obtained. Spatial redundancy is removed by performing orthogonal transformation such as Discrete Cosine Transform (hereinafter referred to as DCT) and quantizing the coefficient obtained by the transformation. By this quantization processing, even if the value of a frequency component having a small coefficient value is set to 0, image quality deterioration can be made inconspicuous if the quantization scale used in the quantization processing is small. Also, motion compensation is used to remove redundancy in the time direction. In motion compensation, for example, in the unit of a macro block of 16 × 16 pixels, the closest region is selected from the reference frame, and the difference value is encoded. When the motion vector amount is small, the difference value is almost 0, and the temporal redundancy can be reduced.

  Here, if the compression rate at the time of encoding is increased, encoding noise is generated and the image quality is deteriorated. Typical coding noise is block noise (also called block distortion, noise that makes the block boundary clearly appear to be tiled) and mosquito noise (noise that appears to have mosquitoes flying around the divided blocks) There is. Unlike analog noise, block noise and mosquito noise are easily noticeable as image quality degradation, and it is necessary to remove these noises.

  In Patent Literature 1, in order to reliably detect a block boundary where block noise occurs or a block where mosquito noise occurs, and to remove those noises, motion compensation prediction for an input image, orthogonal transformation in units of blocks, and The encoded information including the orthogonal transformation coefficient and the motion vector for each block is extracted from the code sequence encoded using each process of quantization by the encoded information extraction step, and the motion vector of each extracted block is extracted. Based on the reference frame, the reference region of each block is obtained by the reference region extraction step, and based on the distribution of each frequency component of the orthogonal transform coefficient relating to the obtained block and the block in the reference frame that overlaps the reference region. The block that generates the coding noise to be removed is detected by the coding noise detection step, and the noise is detected. The boundaries of the pixel blocks of the detected block filtering, detection and removal method of the block noise so as to output an image obtained by removing the block noise has been disclosed as to generate's.

In Non-Patent Document 1, that is, MPEG (moving picture experts group) -4 AVC (Advanced Video Coding), encoding information obtained by encoding a macroblock and a macroblock adjacent to the macroblock are encoded. Describes a method of detecting the intensity of block distortion generated based on the adjacent block coding information obtained by the conversion and performing block noise filter processing based on the detection. That is, the strength of block distortion is obtained as a BS (Boundary strength) value. 0 to 4 are defined as the Bs value. The detection of the Bs value includes the picture type of the macroblock to be encoded, whether the encoding mode is the frame mode or the field mode, the size and direction of the motion vector, the number of reference images, the difference in reference images, and This is performed depending on whether a significant coefficient exists in the block. The Bs value 0 is a case where no block noise removal filter is used, and the Bs value 1 is a case where a small block noise removal characteristic is given. As the Bs value increases, the removal characteristic also increases, and the Bs value 4 is the case where the largest block noise removal characteristic is given. It should be noted that a filter On / Off for executing or not executing the block noise filter insertion processing can be selected.
JP 2001-204029 A ISO / IEC14496-10

  However, in the method disclosed in Non-Patent Document 1, an appropriate Bs value is detected, encoding is performed while performing filter processing using the detected Bs value, and the decoding side is the same as the encoding side. The Bs value is detected, and decoding is performed so as to generate a reference screen subjected to the filter processing based on the Bs value. Here, if there is a difference between the Bs value on the encoding side and the Bs value on the decoding side, the reference images on the encoding side and the decoding side are not the same, and the decoded image includes distortion. If the encoding side transmits the Bs value obtained for each block to the decoding side, the decoding side can perform filtering using the same Bs value, but transmits the Bs value information for each block. Since the coding efficiency deteriorates, the Bs value is obtained based on the coded signal input on the decoding side. The Bs value is a macro for each of the magnitude of the motion vector of the signal to be encoded, the consistency in the vector direction, and whether there is a significant coefficient that is likely to cause block distortion in the coefficient value obtained by orthogonal transformation. The picture type of the block, whether the encoding mode is the frame mode or the field mode, the number of reference images, and the like are obtained by referring to them. Also on the decoding side, the same calculation process as described above is performed and obtained. Even when decoding is performed simply, new decoding distortion will occur if the same Bs value as that used at the time of encoding is obtained for each block by the above arithmetic processing and decoding is not performed. Therefore, the amount of signal processing in the decoding apparatus is large, for example, it is necessary to perform the calculation of the Bs value for all the blocks to be encoded.

  In the method disclosed in Patent Document 1, a filter process that reduces the occurrence of block distortion by obtaining a Bs value at the time of encoding is not performed, and the block distortion reduction process detects a block that gives block distortion exclusively on the decoding side, Block distortion is reduced by filtering the signal of the detected block. Therefore, although the block distortion included in the decoded image is reduced, the high frequency component of the decoded image is also attenuated at the same time, so that the quality of the decoded image is deteriorated. In the case of a decoding device that performs decoding easily, a decoded video signal with an image quality corresponding to the configuration of the device can be obtained, but in the case of a high-performance decoding device, the occurrence of block distortion is reduced and the quality is high. It has not been possible to realize a decoding device that obtains a correct decoded video signal.

  Therefore, the present invention has been made to solve the above-described problems. The encoding device side determines a place where a loop filter needs to be inserted, and the loop filter is inserted by the determination. Encoding to reduce block distortion is performed by increasing the minimum code amount, and the decoding device side reduces the signal processing amount for performing calculations related to the determination of insertion and non-insertion of the loop filter, and performs high-quality decoding. It is an object of the present invention to provide a moving picture coding apparatus and a moving picture coding program capable of obtaining an encoded video signal.

According to a first aspect of the present invention, when an input image that has been divided in advance by a plurality of slices in the horizontal direction is encoded, the block noise generated during decoding of the encoded input image is reduced using a loop filter. In the moving image encoding apparatus, a DCT conversion unit that converts the input image into a coefficient corresponding to a frequency component, a quantization unit that quantizes the coefficient corresponding to the frequency and outputs a quantized value, and the quantization An average value calculation unit that calculates an average quantization value of a macroblock in one slice image of the input image, and a first comparison unit that compares the average quantization value with a predetermined quantization threshold value; A variance value calculation unit that obtains a variance value of the quantized values in the macroblock, a second comparison unit that compares the variance value with a predetermined variance threshold value, and a comparison result in the first comparison unit. Output a signal for turning off the loop filter when the first comparison value is smaller than a predetermined quantization value and / or when the comparison result in the second comparison unit is smaller than a predetermined dispersion value, A signal for turning on the loop filter is output when the comparison result at the comparison unit is larger than a predetermined quantization value and when the comparison result at the second comparison unit is larger than a predetermined variance value. A filter level determination unit that performs the operation, and when a signal that turns on the loop filter is output from the filter level determination unit, from the first macroblock of the sliced input image to the macroblock immediately before the loop filter is turned on If there are consecutive macroblocks that turn off the loop filter, the section of the macroblocks that exist continuously is replaced with a new slice. Determined as the scan interval, to provide a video encoding apparatus is characterized in that and a slice section determining unit for adding a header to the beginning of the new slice interval.
According to a second aspect of the present invention, when an input image that has been divided in advance by a plurality of slices in the horizontal direction is encoded, block noise generated when the encoded input image is decoded is reduced using a loop filter. In the moving image encoding program to be operated, a DCT conversion unit that converts the input image into a coefficient corresponding to a frequency component, a quantization unit that quantizes the coefficient corresponding to the frequency and outputs a quantized value, and the quantum An average value calculation unit that calculates an average quantization value of a macroblock in one slice image of the converted input image, and a first comparison unit that compares the average quantization value and a predetermined quantization threshold A variance value calculation unit that obtains a variance value of quantized values in the macroblock, a second comparison unit that compares the variance value with a predetermined variance threshold, and the first comparison unit When the comparison result is smaller than a predetermined quantization value and / or when the comparison result in the second comparison unit is smaller than a predetermined dispersion value, a signal for turning off the loop filter is output, A signal for turning on the loop filter when a comparison result in the first comparison unit is larger than a predetermined quantization value and a comparison result in the second comparison unit is larger than a predetermined dispersion value. And a macro immediately before turning on the loop filter from the first macroblock of the sliced input image when a signal to turn on the loop filter is outputted from the filter level decision unit If there is a continuous macroblock that turns off the loop filter up to the block, the section of the macroblock that exists continuously is updated. Determined as a slice interval, to provide a moving picture encoding program characterized by having a function to be executed by the control means, and a slice section determining unit for adding a header to the beginning of the new slice interval.

  According to the present invention, an average value calculation unit for obtaining an average quantization value of a macroblock in one slice image of quantized input images, the average quantization value and a predetermined quantization threshold value are obtained. A first comparison unit for comparing, a variance value calculating unit for obtaining a variance value of quantized values in the macroblock, a second comparison unit for comparing the variance value with a predetermined variance threshold, and the first When the comparison result in the comparison unit is smaller than a predetermined quantized value or when the comparison result in the second comparison unit is smaller than a predetermined variance value, a signal for turning off the loop filter is output. When the comparison result in the first comparison unit is larger than a predetermined quantization value, and when the comparison result in the second comparison unit is larger than a predetermined variance value, the loop filter is Output signal to turn on When a signal for turning on the loop filter is output from the filter level determining unit and the filter level determining unit, the loop from the first macroblock of the sliced input image to the macroblock immediately before the loop filter is turned on When there are consecutive macroblocks for which the filter is turned off, the continuous macroblock section is determined as a new slice section, and a slice section determination unit that adds a header to the head of the new slice section Therefore, the coding apparatus determines the location where the loop filter needs to be inserted, and inserts the loop filter according to the determination, thereby reducing the block distortion to the minimum code amount. In addition, the decoding device side decides whether to insert or not insert a loop filter. Reducing the amount of signal processing for performing the calculation according, we can achieve a moving picture coding apparatus and a moving picture coding program as high-quality decoded video signal is obtained.

Hereinafter, a moving picture coding apparatus and a moving picture coding program according to each embodiment of the present invention will be described with reference to FIGS.
FIG. 1 is a block diagram illustrating a schematic configuration example of a moving image encoding apparatus according to an embodiment of the present invention.
FIG. 2 is a diagram illustrating a configuration example of the loop filter On / Off determination unit according to the embodiment of the present invention.
FIG. 3 is a diagram for explaining a loop filter On / Off determination operation according to the embodiment of the present invention.
FIG. 4 is a diagram illustrating a configuration example of the header generation unit of the video encoding device according to the embodiment of the present invention.
FIG. 5 is a flowchart showing the operation of the moving picture coding apparatus according to the embodiment of the present invention.
FIG. 6 is a flowchart showing the operation of the main part of the video encoding apparatus according to the embodiment of the present invention.

  The moving picture coding apparatus and the moving picture coding apparatus according to the moving picture coding program perform coding that reduces block distortion while obtaining a Bs (Boundary strength) value on the coding apparatus side by increasing the minimum code amount. And a video encoding device capable of obtaining a high-quality decoded video signal by reducing the amount of signal processing for performing calculations related to determination of insertion and non-insertion of a loop filter on the decoding device side, and For the purpose of realizing a moving image encoding program, an average value calculation unit for obtaining an average quantization value of a macroblock in one slice image of quantized input images, and the average quantization value are predetermined. A first comparison unit that compares the quantized threshold value, a variance value calculation unit that obtains a variance value of the quantized value in the macroblock, and a first comparison unit that compares the variance value with a predetermined variance threshold value. When the comparison result between the comparison unit and the first comparison unit is smaller than a predetermined quantized value, or when the comparison result at the second comparison unit is smaller than a predetermined variance value, the loop When a signal for turning off the filter is output and the comparison result in the first comparison unit is larger than a predetermined quantization value, and the comparison result in the second comparison unit is larger than a predetermined variance value A filter level determining unit that outputs a signal that turns on the loop filter when the signal is large, and a first macroblock of the sliced input image when a signal that turns on the loop filter is output from the filter level determining unit If there are continuous macroblocks that turn off the loop filter from the macroblock immediately before turning on the loop filter to the macroblock immediately before turning on the loop filter, This is realized by including a slice block determination section that determines a block section of a black block as a new slice section and adds a header to the head of the new slice section.

The configuration of the moving picture coding apparatus will be described.
1 includes an intra prediction unit 11, a motion compensation unit (MC) 12, a motion estimation unit (ME) 13, a subtraction unit 14, a conversion unit 15, a quantization unit 16, Entropy encoding unit 17, memory 18, header generation unit 19, inverse quantization unit 21, inverse transformation unit 22, addition unit 23, decoded image memory 24, loop filter On / Off determination unit 25, loop filter unit 26, and DPB (Decoded Picture Buffer) 27. Connected to the subtractor 14 and the adder 23 is a switch SW1 that has a terminal a and a terminal b and switches an input signal. Connected to the DPB 27 is a switch SW2 having a terminal a and a terminal b for switching an input signal. An image signal is input to the intra prediction unit 11, the motion compensation unit 12, the motion prediction unit 13, and the subtraction unit 14. The header generation unit 19 outputs an encoded signal.
The loop filter On / Off determination unit 25 shown in FIG. 2 includes a variance value calculator 251, a threshold value setter 252, a comparator 253, an average value calculator 254, a threshold value setter 255, a comparator 256, and a filter level determiner 257. Composed.
4 includes a slice header reader 191, a loop filter off information coefficient unit 192, a slice interval determiner 193, a slice information updater 194, and a slice header rewriter 195.

The operation of the moving picture coding apparatus will be described. First, encoding of an intra-frame image (intra image) will be described, and then encoding of an inter-frame image (motion compensated image) performed by reducing block noise will be described.
First, an image of a slice in which a plurality of macroblock images each having 16 pixels in the horizontal direction and the vertical direction are arranged in the horizontal direction is input to one terminal of the subtraction unit 14. An intra image predicted and generated by the intra prediction unit 11 is input to the other terminal of the subtracting unit 14 via the switch SW1, and a difference component between the input image and the intra predicted image is obtained. The conversion unit 15 divides the obtained difference component image into a plurality of block images each having four pixels in the horizontal direction and the vertical direction, and the block image obtained by the division is subjected to DCT (discrete cosine transform) conversion. Thus, coefficient data is obtained. The coefficient data is quantized by the quantization unit 16. The quantized data is variable-length encoded by the entropy encoding unit 17, temporarily stored in the memory 18, and then a compressed encoded signal generated by adding header information described later to the header generation unit 19 is output. At the same time, the data quantized by the quantization unit 16 is input to the inverse quantization unit 21 and inversely quantized. The coefficient data obtained by the inverse quantization is subjected to inverse DCT transform by the inverse transform unit 22. The difference component obtained by the inverse DCT transform is added to the image output from the intra prediction unit 11 by the adding unit 23 to obtain decoded image data. The decoded image data is temporarily stored in the decoded image memory 24 and used as an intra image predicted image related to intra-frame coding in the intra prediction unit 11.

  The image temporarily stored in the decoded image memory 24 is used as an inter-frame prediction image. The block noise generated when decoding the inter-frame prediction image is when the quantized value obtained by DCT transforming each block is large and the quantized value varies greatly between adjacent blocks. The coefficient data obtained by the quantization unit 16 is input to the loop filter On / Off determination unit 25, and the occurrence of block noise during decoding is predicted. The input polarity of the switch SW2 is controlled by the loop filter On / Off determination unit 25. When it is determined that no block noise is generated, the switch SW2 is input to the a side, and when it is determined that block noise is generated, the switch SW2 is input to the b side. From the loop filter On / Off determination unit 25, Bs (Boundary strength) values of 0 to 4 are output according to the magnitude of block noise predicted to occur. When the Bs value is output as 0, the block noise attenuation loop filter is not used, and the switch SW2 is turned on to the a side. When the Bs value is output as 1 to 4, the characteristic of the loop filter unit 26 is set to a characteristic corresponding to the Bs value. The switch SW2 is turned on on the b side. The DPB 27 is input to the decoded image that has been filtered with the characteristic corresponding to the Bs value, and is temporarily stored there.

The motion prediction unit 13 compares the decoded image subjected to the filter processing of the DPB 27 and the input image signal to obtain a motion vector. The motion compensation unit 12 generates a motion prediction image using the motion vector obtained by the motion prediction unit 13 and the decoded image stored in the DPB 27. At the time of encoding the intra-frame encoded image, the switch SW1 is turned on to the b side,
The subtracting unit 14 generates a difference signal component obtained by subtracting the motion prediction image from the input image signal. The difference signal component is DCT-converted by the converter 15 and thereafter in the same manner as that for the intra image. The header generation unit 19 outputs a compressed encoded signal generated by encoding.

  The output compressed encoded signal is decoded by a moving picture decoding apparatus (not shown). The moving picture decoding apparatus is a so-called H.264 standard. The apparatus is configured in the same manner as an apparatus for decoding a signal encoded by H.264. That is, the moving picture decoding apparatus reads the header signal generated by the header generation unit 19, performs entropy decoding based on the read header signal, and obtains a quantized value. The obtained quantized value is subjected to the same processing as that performed by the inverse quantizing unit 21, the inverse transforming unit 22, the adding unit 23, and the decoded image memory 24 to obtain decoded image data. The quantized value is subjected to the same processing as that of the inverse quantization unit 21 to obtain coefficient data, and Bs values of 0 to 4 are obtained by the same processing as the loop filter On / Off determination unit 25. A decoded image that has been subjected to the loop filter process according to the obtained Bs value is obtained.

  Here, the calculation process for obtaining the Bs value needs to be performed for all blocks, and the process is complicated. Therefore, when encoding is performed with the loop filter processing of all macroblocks constituting a slice being set to Off, information in which all the blocks related to the slice are set to Bs = 0 is described in the slice header. A calculation process for obtaining the Bs value on the decoding device side is not necessary. Although there is a method for generating a compressed encoded signal in which Bs value information related to all blocks is described in a slice header, it is not preferable to transmit Bs value information for every block because encoding efficiency deteriorates. Therefore, it is described that the Bs value is 0 in the header of the slice in which the Bs value of all the blocks is 0. In addition, when a macroblock having a Bs value other than 0 is generated and the Bs values of previous macroblocks are continuously 0, a new slice is defined, and the header describes that all blocks have a Bs value of 0. An encoded signal is generated together with the information. On the decoding device side, the number of calculation processing steps for obtaining the Bs value is reduced. The generation of a new slice is executed only when a predetermined number or more of macroblocks having a Bs value of 0 are continuously generated, and the reduction in coding efficiency due to the insertion of a new slice header is reduced.

Next, the loop filter On / Off determination unit 25 will be described in detail with reference to FIG.
First, the coefficient value for each block DCT-transformed by the transform unit 15 and quantized by the quantization unit 16 is input to the average value calculator 254 to calculate the average value, and is also input to the variance value calculator 251. A variance value is determined. Here, when a large variance value is obtained, the image signal level for each block is not uniform, and block noise is likely to occur at the boundary of each block. The threshold value setter 252 stores a threshold value T that is considered to require loop filter processing because occurrence of block noise is predicted. The comparator 253 compares the variance value input from the variance value calculator 251 with the threshold value T input from the threshold value setter 252, and outputs a comparison result relating to execution of the loop filter process when the variance value is larger. The coefficient value relating to each block obtained by the average value calculator 254 is compared with the threshold value set by the threshold value setter 255 by the comparator 256. The comparator 256 outputs a signal for executing loop filter processing when the average value of the coefficients is larger than the threshold value. The filter level determiner 257 generates and outputs a control signal for driving the loop filter 26 and the switch SW2 based on the respective comparison outputs input from the comparator 253 and the comparator 256.

Here, the threshold value set by the threshold value setter 255 is varied according to the size of the variance value output from the variance value calculator 251.
FIG. 3 shows the relationship between the magnitude of the variance value output from the variance value calculator 251 and the threshold value set by the threshold value setter 255. The threshold set by the threshold setter 255 is set to a smaller threshold as the variance value output from the variance value calculator 251 increases. In the figure, the loop filter is turned on at the portion indicated by the halftone dots. Note that in the region where the loop filter is turned on, characteristics for each Bs value (not shown) relating to the characteristics of the loop filter are defined, and a large Bs value is set in the upper left direction.

Generation of a slice header when generating a new slice will be described with reference to FIG.
First, information related to On / Off of the loop filter output from the loop filter On / Off determination unit 25 is input to the loop filter off information coefficient unit 192. There, On / Off for each macroblock is checked based on On / Off information output for each block, and the number of macroblocks in which the loop filter is continuously turned off from the first macroblock of the slice is counted. The Here, when a macroblock in which the loop filter is turned on is detected and the loop filters of a predetermined number of macroblocks or more before that are counted as being continuously off, the slice interval determiner 193 continues. A plurality of macroblocks that are turned off are determined as new slice sections. The slice header reader 191 reads the slice header information output from the entropy encoding unit 17. The slice information updater 194 generates slice header information related to a new slice section based on the slice header information supplied from the slice header reader 191 and the new slice section information supplied from the slice section determiner 193. The In the new slice header, information in which the loop filters of all the macroblocks in the section are turned off is described. The slice header rewrite unit 195 rewrites and outputs the slice header of the compressed encoded signal that has been subjected to variable length coding by the entropy coding unit 17, attached with a slice header, and temporarily stored in the memory 18.

With reference to FIG. 5, the signal flow of the moving picture coding apparatus will be described.
First, an image signal is input in S (step) 71. In S72, it is detected whether or not the input image is an intra (encoded in frame) image. In the case of an intra-frame image, intra prediction is performed in S73. If it is not an intra-frame image, a motion vector is calculated in S74, and motion prediction is performed in S75. The intra-predicted image and the motion-predicted image each having four pixels in the horizontal and vertical directions are subjected to DCT conversion in S76. The coefficient value obtained by DCT conversion is quantized in S77. One of the coefficients obtained in S77 is subjected to variable length coding by the entropy coding unit 17 to generate a compressed coded signal, and the other coefficient is subjected to inverse quantization and inverse DCT in S78. In S79, a decoded image is created. In S81, the Bs value related to the coefficient of the loop filter is obtained based on the variance value of the quantized values and the average value of the quantized values.

  When it is detected in S82 that the loop filter is OFF, the number of blocks or macroblocks in which the filter is OFF is counted in S83, and the process proceeds to S89. When it is detected in S82 that the loop filter is On, after the loop filter processing is performed in S84, the number of macroblocks that are turned off in S85 is a predetermined number, for example, more than half the number of macroblocks constituting the slice. Is detected. When the number of macroblocks is equal to or larger than the predetermined number, header update information for generating a coded signal having a new slice structure is generated in S86, and the slice header is rewritten in S87. After S87 and when the Off count value in S85 is less than or equal to the predetermined number, the count value of the filter Off counted in S83 is cleared. In S89, it is detected whether the image is a reference image or a motion compensated image. If the image is a reference image, the decoded image is held as a reference image in S91. Entropy coding is performed in S92. In S93, it is detected whether there are more input images. If there are more input images, the operation from S71 is repeated.

The loop filter count determination process will be further described with reference to FIG.
After the decoded image is created in S79, the variance of the quantized values of the entire macroblock is obtained in S811. In S812, it is detected whether the quantized variance value is smaller than the threshold value T. If it is smaller, the loop filter flag is turned off in S813. If not, the average quantization value in the macroblock is calculated in S814. In S815, it is detected whether the average quantized value is smaller than the weighted average value obtained by correcting the quantized variance value shown in FIG. If it is smaller, the loop filter flag is turned off in S816. If it is not smaller, the loop filter flag is turned on in S817. After the steps S813, S816, and S817 are completed, the operation of S82 is performed.

  As described above, according to the moving picture coding apparatus of the present embodiment, the average value calculation unit 254 that obtains the average quantization value of the macroblock in one slice image of the quantized input image, and the average A first comparison unit 256 that compares a quantization value with a predetermined quantization threshold; a dispersion value calculation unit 251 that obtains a dispersion value of the quantization value in the macroblock; and the dispersion value and a predetermined value When the comparison result in the second comparison unit 253 for comparing the dispersion threshold and the first comparison unit is smaller than a predetermined quantization value and / or the comparison result in the second comparison unit is predetermined When the variance value is smaller, a signal for turning off the loop filter 26 is output, and when the comparison result in the first comparison unit is larger than a predetermined quantized value, and the comparison in the second comparison unit The result is a predetermined variance value A filter level determining unit 257 that outputs a signal for turning on the loop filter when the signal is larger than the first level of the sliced input image when a signal for turning on the loop filter is output from the filter level determining unit. If there are consecutive macroblocks that turn off the loop filter from the macroblock to the macroblock immediately before the loop filter is turned on, the macroblock is determined as a new slice section by the section of the macroblock that exists continuously. And a slice section determination unit 193 that adds a header to the head of the new slice section. Therefore, the encoding apparatus side determines a place where a loop filter needs to be inserted, and the loop filter is determined by the determination. Minimize coding to reduce block distortion by inserting A video in which a high-quality decoded video signal can be obtained by reducing the amount of signal processing for performing computations related to the determination of insertion and non-insertion of a loop filter on the decoding device side while increasing the signal amount An image encoding device and a moving image encoding program can be realized.

In the moving image encoding apparatus, a compression encoded signal including a block encoded with a loop filter turned on and a compression coded signal coded with all the blocks in the slice turned off. By transmitting the signals separately, the decoding apparatus side does not need to calculate the Bs value for the encoded signals of the slices in which the loop filters of all blocks are turned off, and data processing for decoding The amount can be reduced.
As an application thereof, the same effect can be obtained by attaching a 1-bit flag indicating the macroblock header to the macroblock header in which the loop filters of all the blocks are turned off.

  The present invention can be applied to a moving image encoding apparatus that generates a compressed encoded signal by compressing and encoding an image signal using a loop filter so that block noise does not occur.

It is a block diagram which shows the example of a schematic structure of the moving image encoder which concerns on implementation of this invention. It is a figure which shows the structural example of the loop filter On / Off determination part which concerns on implementation of this invention. It is a figure for demonstrating the loop filter On / Off determination operation | movement which concerns on implementation of this invention. It is a figure which shows the structural example of the header production | generation part of the moving image encoder which concerns on implementation of this invention. It is the figure which showed the operation | movement of the moving image encoder which concerns on implementation of this invention with the flowchart. It is the figure which showed the operation | movement of the principal part of the moving image encoder which concerns on implementation of this invention with the flowchart.

Explanation of symbols

11 Intra Prediction Unit 12 Motion Compensation Unit 13 Motion Prediction Unit 14 Subtraction Unit 15 Conversion Unit 16 Quantization Unit 17 Entropy Coding Unit 18 Memory 19 Header Generation Unit 21 Inverse Quantization Unit 22 Inverse Conversion Unit 23 Addition Unit 24 Decoded Image Memory 25 Loop filter On / Off determination unit 26 Loop filter unit 27 DPB
191 Slice header reader 192 Loop filter off information coefficient unit 193 Slice interval determiner 194 Slice information updater 195 Slice header rewriter 251 Variance calculator 252 Threshold setter 253 Comparator 254 Average value calculator 255 Threshold setter 256 Compare 257 Filter level determiner

Claims (2)

In a video encoding device that reduces block noise generated when decoding an encoded input image using a loop filter when encoding an input image that has been divided in advance by a plurality of slices in the horizontal direction,
A DCT converter for converting the input image into a coefficient corresponding to a frequency component;
A quantization unit that quantizes a coefficient corresponding to the frequency and outputs a quantized value;
An average value calculating unit for obtaining an average quantized value of a macroblock in one slice image of the quantized input image;
A first comparison unit that compares the average quantization value with a predetermined quantization threshold;
A variance value calculation unit for obtaining a variance value of quantized values in the macroblock;
A second comparison unit that compares the dispersion value with a predetermined dispersion threshold;
The loop filter is turned off when a comparison result in the first comparison unit is smaller than a predetermined quantization value and / or when a comparison result in the second comparison unit is smaller than a predetermined variance value. When the comparison result in the first comparison unit is larger than a predetermined quantized value and the comparison result in the second comparison unit is larger than a predetermined variance value A filter level determining unit for outputting a signal for turning on the loop filter;
A macro for turning off the loop filter from the first macroblock of the sliced input image to a macroblock immediately before turning on the loop filter when a signal for turning on the loop filter is output from the filter level determination unit; When the blocks exist continuously, the section of the macroblock that exists continuously is determined as a new slice section, and a slice section determination unit that adds a header to the head of the new slice section;
A moving picture encoding apparatus comprising: Moving picture coding that functions to reduce block noise generated during decoding of the coded input picture by using a loop filter when coding an input picture previously divided by a plurality of slices in the horizontal direction. In the program
A DCT converter for converting the input image into a coefficient corresponding to a frequency component;
A quantization unit that quantizes a coefficient corresponding to the frequency and outputs a quantized value;
An average value calculating unit for obtaining an average quantized value of a macroblock in one slice image of the quantized input image;
A first comparison unit that compares the average quantization value with a predetermined quantization threshold;
A variance value calculation unit for obtaining a variance value of quantized values in the macroblock;
A second comparison unit that compares the dispersion value with a predetermined dispersion threshold;
The loop filter is turned off when a comparison result in the first comparison unit is smaller than a predetermined quantization value and / or when a comparison result in the second comparison unit is smaller than a predetermined variance value. When the comparison result in the first comparison unit is larger than a predetermined quantized value and the comparison result in the second comparison unit is larger than a predetermined variance value A filter level determining unit for outputting a signal for turning on the loop filter;
A macro for turning off the loop filter from the first macroblock of the sliced input image to a macroblock immediately before turning on the loop filter when a signal for turning on the loop filter is output from the filter level determination unit; When the blocks exist continuously, the section of the macroblock that exists continuously is determined as a new slice section, and a slice section determination unit that adds a header to the head of the new slice section;
A moving picture encoding program characterized by comprising a function for causing the control means to execute.

Images