JP2009017509A - Apparatus and method for decoding images - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To configure a post filter with small circuit scale of and capable of easily responding to an adaptive treatment. <P>SOLUTION: An image decoder 100 is equipped with a variable-length decoder 2 for decoding encoded data into variable length; an inverse-quantization inverse orthogonal transformer 13 for subjecting the output of the variable-length decoder 2 into inverse quantization and an inverse orthogonal transformation, to generate a differential pixel value in 11-bit accuracy; an image-data converter 11 for converting the differential pixel value in 11-bit accuracy to the value into 9-bit accuracy; a memory 5 for storing a reference image in 8-bit accuracy; an adder 4 for adding the reference image and the differential pixel value in 9-bit accuracy, to store them in the memory 5 as the reference image; an adder 14 for adding the reference image and the differential pixel value in 11-bit accuracy stored in the memory 5, to generate macroblock-unit image data in 10-bit accuracy; a post filter 20 for applying filtering processing on the macroblock-unit image data; and a memory 15 for storing the filtered image data to output them in field units and frame units. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an image decoding apparatus and an image decoding method, and in particular, is an image decoding apparatus that conforms to an image encoding standard that defines input / output of image data with a first bit accuracy, The present invention relates to an image decoding apparatus that decodes encoded data obtained by encoding image data with second bit accuracy higher than bit accuracy into image data with second bit accuracy.

  In recent years, a service for distributing a compressed stream obtained by encoding a moving image obtained by a camera via terrestrial broadcasting, satellite broadcasting, the Internet, or the like has become widespread. In addition, digital video cameras and digital video recorders that record a compressed stream obtained by encoding a moving image obtained by a camera on a recording medium as a digital signal have become widespread. Examples of methods used for such encoding processing include MPEG2 (ISO / IEC 13818-2), MPEG4 (ISO / IEC 14496-2), and H.264. There is an image coding method using motion detection and motion compensation such as H.264 (ITU-T Rec. H.264). In these image encoding methods, image encoding is performed by using temporal correlation between frames or fields in a moving image in addition to correlation between adjacent pixels in a frame.

  As an image encoding device that performs such image encoding, for example, an image decoding device described in Non-Patent Document 1 is known.

Hereinafter, a conventional image encoding device will be described.
FIG. 7 is a block diagram showing a configuration of a conventional image encoding device.

  7 includes an image input unit 51, a subtracting unit 52, an orthogonal transform quantizing unit 53, a variable length coding unit 54, a code output unit 55, and an inverse quantization inverse. An orthogonal transformation unit 56, an addition unit 57, a memory 58, and a motion compensation unit 59 are provided.

  The image input unit 51 rearranges the input image data with 8-bit accuracy in the frame order. Further, the image input unit 51 cuts out input image data in units of macroblocks used for motion compensation processing and orthogonal transformation processing from input image data in units of frames.

  The subtractor 52 subtracts the 8-bit accuracy predicted image data output from the motion compensation unit 59 from the input image data cut out by the image input unit 51 to generate a 9-bit accuracy difference pixel value.

  The orthogonal transform quantization unit 53 generates a quantized orthogonal transform coefficient by performing orthogonal transform processing and quantization processing on the 9-bit precision difference pixel value generated by the subtraction unit 52.

  The variable length coding unit 54 performs variable length coding on the quantized orthogonal transform coefficient generated by the orthogonal transform quantization unit 53.

  The code output unit 55 outputs the encoded data subjected to the variable length encoding by the variable length encoding unit 54 as a compressed stream. The encoded data output from the code output unit 55 is used for transmission and recording.

  The inverse quantization inverse orthogonal transform unit 56 generates a difference pixel value with 9-bit accuracy by performing inverse quantization processing and inverse orthogonal transform processing on the quantized orthogonal transform coefficient generated by the orthogonal transform quantization unit 53. To do.

  The adding unit 57 adds the 9-bit accuracy difference pixel value generated by the inverse quantization inverse orthogonal transform unit 56 and the 8-bit accuracy predicted image data output from the motion compensation unit 59, thereby obtaining 8-bit accuracy. Generate a local decoded image in units of macroblocks.

  The memory 58 stores the local decoded image generated by the adding unit 57. The memory 58 stores a plurality of locally decoded images in units of macroblocks as reference image frames in units of frames.

  The motion compensation unit 59 uses the reference image frames stored in the memory 58 to generate predicted image data in units of macro blocks with an 8-bit accuracy for an image of a frame input later.

  As described above, the conventional image encoding device 400 encodes input image data with 8-bit accuracy and outputs encoded data.

Next, a conventional image decoding apparatus will be described.
FIG. 8 is a block diagram showing a configuration of a conventional image decoding apparatus.

  The image decoding apparatus shown in FIG. 8 decodes the encoded data generated by the image encoding apparatus 400 shown in FIG. The image decoding apparatus 500 includes a code input unit 1, a variable length decoding unit 2, an inverse quantization inverse orthogonal transform unit 3, an addition unit 4, a memory 5, a motion compensation unit 6, a post filter 21, The image output unit 7 is provided.

  The code input unit 1 accumulates code data output from the image encoding device 400 shown in FIG. The code input unit 1 also performs header separation and analysis on the code data output from the image encoding device 400.

  The variable length decoding unit 2 performs variable length decoding on the encoded data that has been subjected to header separation and analysis by the code input unit 1, and generates quantized orthogonal transform coefficients.

  The inverse quantization inverse orthogonal transform unit 3 generates a difference pixel value with 9-bit accuracy by performing an inverse quantization process and an inverse orthogonal transform process on the quantized orthogonal transform coefficient generated by the variable length decoding unit 2. To do.

  The adder 4 adds the 9-bit accuracy difference pixel value generated by the inverse quantization inverse orthogonal transform unit 3 and the 8-bit accuracy predicted image data output from the motion compensation unit 6, thereby obtaining 8-bit accuracy. Decode image data in units of macro blocks is generated.

  The memory 5 stores the decoded image data generated by the adding unit 4. The memory 5 stores decoded image data in units of a plurality of macroblocks as decoded images in units of frames or fields.

  The motion compensation unit 6 uses the decoded image stored in the memory 5 to generate prediction image data in units of macroblocks with an 8-bit accuracy for an image of a frame input later.

  Here, the loop-shaped process including the adder 4, the memory 5, and the motion compensator 6 includes a loop-shaped process including the adder 57, the memory 58, and the motion compensator 59 of the image encoding device 400, an input signal, Processing content and bit accuracy are the same. As a result, the local decoded image generated by the adding unit 57 matches the decoded image generated by the adding unit 4.

  The post filter 21 performs a filter process for reducing noise generated by the encoding process on the decoded image in frame units or field units stored in the memory 5.

  The image output unit 7 outputs the decoded image whose noise has been reduced by the post filter 21 as 8-bit precision image data.

  As described above, the conventional image decoding apparatus 500 decodes encoded data obtained by encoding 8-bit precision image data and outputs 8-bit precision image data.

  Further, as a conventional image encoding device and image decoding device, an image encoding device and an image for performing encoding processing and decoding processing of 10-bit accuracy image data while being compatible with an 8-bit accuracy image encoding method A decoding device is known (for example, refer to Patent Document 1). The image encoding device and the image decoding device described in Patent Document 1 comply with the image encoding standard and can perform encoding and decoding with high image quality.

  FIG. 9 is a block diagram showing a configuration of a conventional image encoding device described in Patent Document 1. In FIG. In addition, the same code | symbol is attached | subjected to the element similar to FIG. 7, and detailed description is abbreviate | omitted.

  9 includes an image input unit 61, a subtraction unit 62, an orthogonal transform quantization unit 63, a variable length coding unit 54, a code output unit 55, and an inverse quantization inverse orthogonal transform. A unit 56, an adder 57, a memory 58, and a motion compensation unit 59.

  The image input unit 61 rearranges the input image data with 10-bit accuracy in the frame order. Further, the image input unit 61 cuts out input image data in units of macroblocks used for motion compensation processing and orthogonal transformation processing from input image data in units of frames.

  The subtractor 62 subtracts the 8-bit accuracy predicted image data output from the motion compensation unit 59 from the input image data cut out by the image input unit 61 to generate an 11-bit accuracy difference pixel value.

  The orthogonal transform quantization unit 63 generates a quantized orthogonal transform coefficient by performing orthogonal transform processing and quantization processing on the 11-bit precision difference pixel value generated by the subtraction unit 62.

  The variable length coding unit 54 performs variable length coding on the quantized orthogonal transform coefficient generated by the orthogonal transform quantization unit 63.

  The code output unit 55 outputs the encoded data subjected to the variable length encoding by the variable length encoding unit 54 as a compressed stream. The encoded data output from the code output unit 55 is used for transmission and recording.

  The inverse quantization inverse orthogonal transform unit 56 generates a difference pixel value with 9-bit accuracy by performing an inverse quantization process and an inverse orthogonal transform process on the quantized orthogonal transform coefficient generated by the orthogonal transform quantization unit 63. To do.

  As described above, the conventional image encoding apparatus 600 encodes 10-bit precision image data using the 8-bit precision predicted image data, and outputs the encoded data. As a result, the loop-like process including the adder 57, the memory 58, and the motion compensation unit 59 of the image encoding device 600 is the loop-like process including the adder 4, the memory 5, and the motion compensation unit 6 of the image decoding device 500. The processing, input signal, processing content, and bit accuracy are the same. Therefore, the local decoded image generated by the adding unit 57 matches the decoded image generated by the adding unit 4. Thereby, the encoded data encoded by the image encoding device 600 can be decoded by the conventional image decoding device 500 shown in FIG. That is, the image encoding device 600 can encode 10-bit precision image data in accordance with an image coding standard that defines input / output of 8-bit precision image data.

  Next, a conventional image decoding apparatus that decodes encoded data encoded by the image encoding apparatus 600 will be described.

  FIG. 10 is a block diagram illustrating a configuration of a conventional image decoding apparatus that decodes encoded data encoded by the image encoding apparatus 600. Although there is no description about the post filter in Patent Document 1, the image decoding apparatus described in Patent Document 1 performs the filtering process in units of frames or fields similar to the conventional image decoding apparatus 500 shown in FIG. FIG. 10 shows a configuration when a post filter is provided.

  An image decoding apparatus 700 illustrated in FIG. 10 includes a code input unit 1, a variable length decoding unit 2, an inverse quantization inverse orthogonal transform unit 13, an image data conversion unit 11, an addition unit 4, a memory 5, and the like. A motion compensation unit 6, an addition unit 14, a memory 15, a post filter 22, and an image output unit 17. In addition, the same code | symbol is attached | subjected to the element similar to FIG. 8, and detailed description is abbreviate | omitted.

  The code input unit 1 accumulates code data output from the image encoding device 600 shown in FIG. The code input unit 1 also performs header separation and analysis on the code data output from the image encoding device 600.

  The variable length decoding unit 2 performs variable length decoding on the encoded data that has been subjected to header separation and analysis by the code input unit 1, and generates quantized orthogonal transform coefficients.

  The inverse quantization inverse orthogonal transform unit 13 performs an inverse quantization process and an inverse orthogonal transform process on the quantized orthogonal transform coefficient generated by the variable length decoding unit 2, thereby generating a difference pixel value with 11-bit accuracy. To do.

  The image data converter 11 converts the 11-bit precision difference pixel value generated by the inverse quantization inverse orthogonal transform section 13 into a 9-bit precision difference pixel value.

  The adder 4 adds the 9-bit precision difference pixel value converted by the image data converter 11 and the 8-bit precision predicted image data output from the motion compensator 6, thereby providing an 8-bit precision macroblock unit. The decoded image data is generated.

  The memory 5 stores the decoded image data generated by the adding unit 4. The memory 5 stores decoded image data in units of a plurality of macroblocks as reference images in units of frames or fields.

  The motion compensation unit 6 uses the reference image stored in the memory 5 to generate predicted image data in units of macroblocks with an 8-bit accuracy for a frame image to be input later.

  Here, the loop-shaped process including the adder 4, the memory 5, and the motion compensator 6 includes a loop-shaped process including the adder 57, the memory 58, and the motion compensator 59 of the image encoding device 600, an input signal, Processing content and bit accuracy are the same. As a result, the local decoded image generated by the adding unit 57 matches the decoded image generated by the adding unit 4.

  The adder 14 adds the difference pixel value with 11-bit accuracy generated by the inverse quantization inverse orthogonal transform unit 13 and the predicted image data with 8-bit accuracy output from the motion compensation unit 6, thereby adding 10-bit accuracy. Decode image data in units of macro blocks is generated.

  The memory 15 stores the decoded image data generated by the adding unit 14. The memory 15 stores the decoded image data in units of a plurality of macro blocks as decoded images in units of frames or fields.

  The post filter 22 performs a filter process for reducing noise generated by the encoding process on the decoded image in frame units or field units stored in the memory 15.

  The image output unit 17 outputs the decoded image whose noise is reduced by the post filter 22 as 10-bit precision image data.

As described above, the conventional image decoding apparatus 700 decodes encoded data in which 10-bit precision image data is encoded in accordance with an image encoding standard that defines input / output of 8-bit precision image data. 10-bit precision image data is output.
ASCII "Latest MPEG Textbook" p138, 1994 JP 2000-175200 A

  However, in the conventional image decoding apparatus 500 shown in FIG. 8, the post filter 21 performs a filtering process on the decoded image in field units or frame units after being stored in the memory 5. When the filter process is performed on the decoded image in units of macroblocks before being stored in the memory 5, the filter process is included in the loop-shaped process including the adder 4, the memory 5, and the motion compensation unit 6. For this reason, the loop processing of the image decoding device 500 and the loop processing including the adder 57, the memory 58, and the motion compensation unit 59 of the image encoding device 400 are not the same. This makes it noncompliant with the standard. Therefore, when the post filter 21 performs vertical filter processing, it is necessary to incorporate a line memory in the field or frame in the post filter 21, and there is a problem that the circuit scale of the post filter 21 becomes large.

  Similarly, in the conventional image decoding apparatus 700 shown in FIG. 10, the post filter 22 performs a filtering process on the decoded image in field units or frame units stored in the memory 15. As a result, there is a problem that the circuit scale of the post filter 22 is increased. Furthermore, since the post filter 22 needs to perform a filtering process on the inputted 10-bit precision decoded image, the circuit scale becomes larger than the post filter 21 of the image decoding apparatus 500 shown in FIG.

  The quantization parameter is given to a macro block unit or a further subdivided block unit. Therefore, in the post-filter 22, when adaptive filter processing is performed using parameters used for quantization, processing is performed in units of macroblocks or blocks for images in units of fields or frames to be processed. Therefore, the control becomes complicated.

  Therefore, the present invention provides an image decoding apparatus and an image decoding method capable of reducing the circuit scale of a post filter and realizing an adaptive filter process using a quantization parameter with simpler control. The purpose is to provide.

  In order to achieve the above object, an image decoding apparatus according to the present invention includes variable length decoding means for generating quantized orthogonal transform coefficients by variable length decoding encoded data, and the quantized orthogonal transform coefficients. Is subjected to inverse quantization and inverse orthogonal transform to generate a first difference pixel value having a bit accuracy and a second difference bit value having a higher accuracy than the first bit accuracy. Means, first memory means for holding one or more reference images of the first bit precision, and of the one or more reference images held in the first memory means, by the inverse quantization inverse orthogonal transform means A first addition unit that adds the reference image corresponding to the generated difference pixel value of the first bit accuracy and the difference pixel value and causes the first memory unit to store the reference image as a reference image; Retained in memory means The reference image corresponding to the difference pixel value of the second bit precision generated by the inverse quantization inverse orthogonal transform unit among the one or more reference images and the difference pixel value are added, thereby adding the difference pixel value. Second addition means for generating second bit precision macroblock unit image data, and filter means for performing filter processing for reducing noise in the macroblock unit image data generated by the second addition means And second memory means for accumulating the image data in units of macroblocks filtered by the filter means and outputting the image data in field units or frame units with the second bit accuracy.

  According to this configuration, the filter means performs a filtering process on the image data in units of macroblocks. As a result, the image decoding apparatus according to the present invention can reduce the circuit scale of the filter means (post filter) as compared with the conventional image decoding apparatus that performs filter processing on image data in frame units or field units. Can do. For example, when performing filter processing in a direction in which pixel signals of pixels are sequentially read (for example, the horizontal direction) and a direction orthogonal to the direction (for example, the vertical direction), the conventional image decoding apparatus uses pixels in the horizontal direction of the frame. Whereas several delay circuits are required, the image decoding apparatus according to the present invention may be provided with delay circuits corresponding to the number of pixels in the horizontal direction of the macroblock.

  In addition, when the filter means dequantizes the quantized orthogonal transform coefficient corresponding to the macroblock unit image data, according to the quantization parameter used in the inverse quantization inverse orthogonal transform means, Filtering characteristics applied to the image data of the macroblock may be changed.

  According to this configuration, the image decoding apparatus according to the present invention determines characteristics of filter processing for image data input in units of macroblocks using quantization parameters determined in units of macroblocks. Thereby, the image decoding apparatus according to the present invention can realize adaptive filter processing using a quantization parameter with simpler control.

  Further, the filter means may perform filter processing with different filter characteristics corresponding to the magnitude of a quantization step indicating a quantization interval included in the quantization parameter.

  According to this configuration, for example, when the quantization step is rough and the encoding accuracy is low, the filter means performs the filtering process with the gain of the high frequency component greatly reduced, thereby reducing noise generated during encoding. If the quantization step is fine and the coding accuracy is high, the adaptive processing can be performed such that high-accuracy image data is output as it is by performing filter processing that does not significantly reduce the gain of the high-frequency component. . That is, the image decoding apparatus according to the present invention can perform an optimum filter process according to the encoding accuracy.

  Further, the inverse quantization inverse orthogonal transform means performs first quantization inverse transform that generates the difference pixel value with the second bit accuracy by performing inverse quantization and inverse orthogonal transform on the quantized orthogonal transform coefficient. The first bit precision difference pixel value generated by the orthogonal transformation means and the first inverse quantization inverse orthogonal transformation means is converted into the first bit precision difference pixel value, thereby generating the first bit precision difference pixel value. Conversion means for generating the difference pixel value with a bit accuracy of.

  According to this configuration, the conversion unit can form the first pixel precision difference pixel value by a simple process such as rounding the lower bits of the second bit precision difference pixel value with high bit precision. . Therefore, the image decoding apparatus according to the present invention can easily generate the difference pixel values having the first bit accuracy and the second bit accuracy.

  In addition, the inverse quantization inverse orthogonal transform unit performs first quantization inverse transform that generates the difference pixel value with the first bit precision by performing inverse quantization and inverse orthogonal transform on the quantized orthogonal transform coefficient. An orthogonal transform unit; and a second inverse quantization inverse orthogonal transform unit that generates the difference pixel value with the second bit precision by performing inverse quantization and inverse orthogonal transform on the quantized orthogonal transform coefficient. Also good.

  The image decoding apparatus further includes a deblocking filter for reducing noise at a macroblock boundary of the image data added by the first adding unit, and the first memory unit uses the deblocking filter to reduce noise. The image data with reduced image may be stored as a reference image.

  According to this configuration, the image decoding apparatus according to the present invention is an H.264 standard. H.264 standard image data can be decoded.

  The image decoding method according to the present invention also includes a variable length decoding step for generating a quantized orthogonal transform coefficient by variable length decoding encoded data, and inverse quantization and inverse processing of the quantized orthogonal transform coefficient. An inverse quantization inverse orthogonal transform step of generating a difference pixel value having a first bit accuracy and a difference pixel value having a second bit accuracy higher than the first bit accuracy by performing orthogonal transform; The first accumulating step of holding one or more reference images of bit accuracy in the first memory means and the inverse quantization inverse orthogonal transform step of the one or more reference images held in the first memory means A first addition step of adding the reference image corresponding to the difference pixel value of the first bit precision and the difference pixel value and causing the first memory means to hold the reference image as a reference image; and the first memory means Held in Adding the reference image corresponding to the difference pixel value of the second bit precision generated in the inverse quantization inverse orthogonal transform step among the one or more reference images, and the difference pixel value, A second adding step for generating image data in units of macroblocks of the second bit precision, and a filter for performing filter processing for reducing noise on the image data in units of macroblocks generated in the second adding step A second accumulating step for accumulating the image data in units of macroblocks filtered in the filtering step in a second memory means, and the second bit precision image accumulated in the second memory means An output step of outputting data in field units or frame units.

  According to this, filter processing is performed on image data in units of macroblocks in the filter step. As a result, the image decoding method according to the present invention has a circuit scale of filter means (post filter) for performing filter processing, as compared with a conventional image decoding method for performing filter processing on image data in units of frames or fields. Can be reduced. For example, when performing filter processing in a direction in which pixel signals of pixels are sequentially read (for example, the horizontal direction) and a direction orthogonal to the direction (for example, the vertical direction), in the conventional image decoding method, the filter means Whereas a delay circuit corresponding to the number of pixels in the horizontal direction is required, in the image decoding method according to the present invention, the filter means may include a delay circuit corresponding to the number of pixels in the horizontal direction of the macroblock.

  The integrated circuit according to the present invention includes variable length decoding means for generating quantized orthogonal transform coefficients by variable length decoding encoded data, and inverse quantization and inverse orthogonal transform of the quantized orthogonal transform coefficients. An inverse-quantized inverse orthogonal transform means for generating a difference pixel value having a first bit precision and a difference pixel value having a second bit precision higher than the first bit precision, and the first bit First memory means for holding one or more reference images with accuracy, and the first bit generated by the inverse quantization inverse orthogonal transform means among the one or more reference images held in the first memory means A reference image corresponding to the difference pixel value of accuracy and the difference pixel value are added together, and the first addition unit causes the first memory unit to hold the reference image as the reference image, and the first memory unit holds the first image. Of the above reference images The second bit-accuracy macroblock is obtained by adding a reference image corresponding to the second-bit-accuracy difference pixel value generated by the inverse quantization inverse orthogonal transform means and the difference-pixel value. Second adding means for generating image data in units, filter means for performing filter processing for reducing noise on the image data in units of macroblocks generated by the second adding means, and filter processing by the filter means And second memory means for storing the image data in units of macroblocks and outputting the image data in field units or frame units with the second bit precision.

  According to this configuration, the filter means performs a filtering process on the image data in units of macroblocks. As a result, the integrated circuit according to the present invention can reduce the circuit scale of the filter means (post filter) as compared with the conventional integrated circuit that performs the filtering process on the image data in frame units or field units. For example, when performing filter processing in a direction in which pixel signals of pixels are sequentially read (for example, the horizontal direction) and a direction orthogonal to the direction (for example, the vertical direction), in a conventional integrated circuit, the number of pixels in the horizontal direction of the frame On the other hand, the integrated circuit according to the present invention only needs to include a delay circuit for the number of pixels in the horizontal direction of the macroblock.

  The present invention can be realized not only as such an image decoding apparatus and image decoding method, but also as a program for causing a computer to execute characteristic steps included in the image decoding method. You can also. Needless to say, such a program can be distributed via a recording medium such as a CD-ROM or a transmission medium such as the Internet.

  The present invention provides an image decoding apparatus and an image decoding method capable of reducing the circuit scale of a post filter and realizing an adaptive filter process using a quantization parameter with simpler control. be able to.

  Hereinafter, embodiments of an image decoding apparatus according to the present invention will be described in detail with reference to the drawings.

(Embodiment 1)
The image decoding apparatus according to Embodiment 1 of the present invention includes a post filter that performs filter processing on a decoded image in units of macroblocks. As a result, the circuit scale of the post filter can be reduced.

First, the configuration of the image decoding apparatus according to the embodiment of the present invention will be described.
FIG. 1 is a block diagram showing a configuration of an image decoding apparatus according to Embodiment 1 of the present invention. In addition, the same code | symbol is attached | subjected to the element similar to FIG.

  An image decoding apparatus 100 shown in FIG. 1 is an image decoding apparatus compliant with an image encoding standard (for example, MPEG2) in which input / output of 8-bit precision image data is defined. Code data obtained by encoding is decoded. In addition, the image decoding apparatus 100 decodes encoded data obtained by encoding 10-bit precision image data while conforming to an image encoding standard that defines input / output of 8-bit precision image data. Outputs 10-bit precision image data.

  For example, the image decoding apparatus 100 decodes the encoded data encoded by the image encoding apparatus 600 shown in FIG. 9 into 10-bit precision image data and outputs the decoded image data. At the same time, the image decoding apparatus 100 outputs image data with 10-bit accuracy even when 8-bit accuracy image data is input. For example, the image decoding apparatus 100 decodes the encoded data encoded by the image encoding apparatus 400 shown in FIG. 7 into 10-bit precision image data and outputs the decoded image data.

  The image decoding apparatus 100 includes a code input unit 1, a variable length decoding unit 2, an inverse quantization inverse orthogonal transform unit 13, an image data conversion unit 11, an adder 4, a memory 5, and a motion compensation unit. 6, an adder 14, a memory 15, a post filter 20, and an image output unit 17.

  The code input unit 1 accumulates code data included in the compressed stream output from the image encoding device 600 shown in FIG. The code input unit 1 also performs header separation and analysis on the code data output from the image encoding device 600.

  The variable length decoding unit 2 generates quantized orthogonal transform coefficients in units of macroblocks by variable length decoding the encoded data that has been subjected to header separation and analysis by the code input unit 1.

  The inverse quantization inverse orthogonal transform unit 13 performs an inverse quantization process and an inverse orthogonal transform process on the quantized orthogonal transform coefficient generated by the variable length decoding unit 2 to thereby obtain a difference in macroblock units with 11-bit accuracy. Generate pixel values.

  The image data conversion unit 11 converts the difference pixel value with 11-bit accuracy generated by the inverse quantization inverse orthogonal transformation unit 13 into a difference pixel value with 9-bit accuracy, so that a difference pixel in macroblock units with 9-bit accuracy is obtained. Generate a value. Specifically, the image data conversion unit 11 generates a 9-bit precision difference pixel value by rounding the lower 2 bits of the 11-bit precision difference pixel value.

  The motion compensation unit 6 uses the reference image stored in the memory 5 to generate predicted image data in units of macroblocks with an 8-bit accuracy for a frame image input later. Specifically, the motion compensation unit 6 uses the 8-bit corresponding to the macroblock of the difference pixel value from the reference image frame corresponding to the difference pixel value of 9-bit accuracy generated by the inverse quantization inverse orthogonal transform unit 13. Accurate predicted image data is generated and output to the adders 4 and 14.

  The adder 4 adds the 9-bit precision difference pixel value generated by the image data converter 11 and the 8-bit precision predicted image data output from the motion compensator 6, thereby providing an 8-bit precision macroblock unit. Decoded image data, which is the image data of, is generated.

  The memory 5 stores the decoded image data generated by the adding unit 4. The memory 5 stores decoded image data in units of a plurality of macro blocks and holds them as reference images in units of frames or fields.

  Here, the loop-shaped process including the adder 4, the memory 5, and the motion compensator 6 includes a loop-shaped process including the adder 57, the memory 58, and the motion compensator 59 of the image encoding device 600, an input signal, Processing content and bit accuracy are the same. As a result, the local decoded image generated by the adding unit 57 and the decoded image data generated by the adding unit 4 match.

  The adder 14 adds the difference pixel value with 11-bit accuracy generated by the inverse quantization inverse orthogonal transform unit 13 and the predicted image data with 8-bit accuracy output from the motion compensation unit 6, thereby adding 10-bit accuracy. Decode image data in units of macro blocks is generated.

  The post filter 20 performs a filtering process for reducing noise generated by the encoding process on the decoded image data generated by the adding unit 14. Specifically, the post filter 20 performs processing to reduce high frequency components included in the decoded image data. For example, the post filter 20 calculates an average value of a plurality of pixels for each of the vertical direction and the horizontal direction of the decoded image data, and performs weighting to reduce high-frequency components.

  Further, the post-filter 20 decodes a predetermined macroblock in the inverse quantization inverse orthogonal transform unit 13 according to the quantization parameter used for inverse quantization of the quantized orthogonal transform coefficient corresponding to the predetermined macroblock. Change the filtering characteristics of the image data.

  The memory 15 stores the decoded image data filtered by the post filter 20. The memory 15 stores the decoded image data in units of a plurality of macro blocks as decoded images in units of frames or fields.

  The image output unit 17 outputs the decoded image in frame units or field units stored in the memory 15 as image data with 10-bit accuracy.

  Next, detailed operation of the post filter 20 of the image decoding apparatus 100 according to Embodiment 1 of the present invention will be described.

FIG. 2 is a diagram illustrating a processing image of the post filter.
2A shows a three-tap vertical filter in which the post filter 22 of the conventional image decoding apparatus 700 shown in FIG. It is a figure which shows the processing image at the time of multiplying. In FIG. 2A, a two-dimensional image is used, but actually, pixel signals that are signals corresponding to image data of each pixel of the decoded image are sequentially input. Specifically, the pixel signal of the pixel in the right direction is read from the memory 15 in order from the pixel signal of the pixel 211 at the upper left end of the decoded image frame 201 shown in FIG. When the pixel signals up to the rightmost pixel of the decoded image frame 201 are read, the pixel signals of the rightward pixels are read in order from the leftmost pixel 212 of the next line.

  As shown in FIG. 2A, the distance between the pixels on which the vertical filter processing is performed by the post filter 22 is separated by the number of pixels in the horizontal direction of the frame. For example, the number of pixels between the pixel 211 and the pixel 212 is separated by the number of pixels Nfh in the horizontal direction of the decoded image frame 201. Thus, the post filter 22 needs to include delay circuits for the number of pixels Nfh. Note that the operation of the post filter 21 of the conventional image decoding apparatus 500 shown in FIG. 8 is the same as the operation of the post filter 22.

  FIG. 2B shows that the post filter 20 of the image decoding apparatus 100 according to Embodiment 1 of the present invention has 3 taps with respect to the decoded image data 221 in units of macroblocks before being stored in the memory 15. It is a figure which shows the process image in the case of applying a vertical filter. In FIG. 2B, the image is expressed as a two-dimensional image. Actually, however, the pixels in the right direction in order from the pixel signal of the pixel 211 at the upper left corner of the decoded image data 221 shown in FIG. Are input to the post filter 20. When pixel signals up to the right end pixel of the decoded image data 221 are input, pixel signals of pixels in the right direction are input in order from the left end pixel 212 of the next line. In this way, when the pixel signals of all the pixels up to the lower right end of the decoded image data 221 are input, the pixels in the right direction are sequentially sequentially from the pixel signal of the upper left pixel of the decoded image data 222 adjacent to the left. The pixel signal is input.

  As shown in FIG. 2B, the distance between the pixels on which the vertical filter processing is performed by the post filter 20 is separated by the number of pixels in the horizontal direction of the macroblock. For example, the number of pixels between the pixel 211 and the pixel 212 is separated by the number of pixels Nmbh in the horizontal direction of the decoded image data 221. As a result, the post filter 20 only needs to have a delay circuit for the number of pixels Nmbn, so that the circuit scale of the delay circuit can be reduced as compared with the conventional post filter 22.

  That is, in the conventional image decoding apparatus 700, when 3-tap vertical filter processing is performed on the pixels 211, 212, and 213 shown in FIG. 2, two lines of pixels in one frame + 1 pixel, that is, 2 × Nfh + 1 It is necessary to hold pixel image data. On the other hand, in the image decoding apparatus 100 according to Embodiment 1 of the present invention, it is only necessary to hold image data of 2 lines of pixels + 1 pixel, that is, 2 × Nmb + 1 pixels, of one macroblock. Therefore, the image decoding apparatus 100 according to Embodiment 1 of the present invention can reduce the capacity of the memory used for the filter processing.

  For example, MPEG2 and H.264 commonly used for high-definition images. In the case of H.264 encoding and decoding, the number of horizontal pixels Nfh in one frame is 1920 pixels only in the effective portion, whereas the number of horizontal pixels Nmbh in one macroblock is 16 pixels. The scale of the delay circuit and the memory of the filter 20 can be significantly reduced from the scale of the delay circuit and the memory of the post filter 22.

  Also, the quantization parameter at the time of encoding is determined on a macroblock basis. Therefore, in the conventional image decoding apparatus 700, since the pixel signals are sequentially input in the horizontal direction of the frame, the pixel signals input during the filtering process straddle a plurality of macroblocks, and an adaptive filter It is difficult to process. On the other hand, in the image decoding apparatus 100 according to Embodiment 1 of the present invention, since pixel signals are input in units of macroblocks, adaptive filter processing in accordance with quantization parameters can be realized by simple control.

  Hereinafter, the details of adaptive filter processing in accordance with the quantization parameter of the post filter 20 will be described.

FIG. 3 is a flowchart showing the flow of filter processing of the post filter 20.
As shown in FIG. 3, first, the post filter 20 acquires the quantization parameter 30 used for inverse quantization in the inverse quantization inverse orthogonal transform unit 13 of the decoded image data to be subjected to filter processing (S101). That is, the post filter 20 acquires the quantization parameter used for quantization in the orthogonal transform quantization unit 63 of the image encoding device 600. The quantization parameter is included in the encoded data input to the code input unit 1 and is used for inverse quantization in the inverse quantization inverse orthogonal transform unit 13.

  Next, the post filter 20 performs a filtering process according to the acquired quantization parameter 30. Specifically, when the quantization step included in the quantization parameter 30 is coarser than a predetermined value (Yes in S102), the post filter 20 performs a filtering process on the decoded image data with the gain of the high frequency component greatly reduced ( S103). Here, the quantization step indicates a quantization interval. That is, when the quantization step is coarse, the encoding accuracy is low, and when the quantization step is fine, the encoding accuracy is high. For example, the post filter 20 selects a coefficient that lowers the cutoff frequency of the low-pass filter and performs filter processing. Specifically, the post filter 20 uses a weighting factor of 1: 2: 1 for three consecutive pixels in the vertical direction in the vertical three-tap filter process. Thereby, when the encoding accuracy is low, the encoding noise can be reduced by the filter processing in which the gain of the high frequency component is greatly reduced.

  If the quantization step included in the quantization parameter 30 is finer than a predetermined value (No in S102), the post filter 20 uses a filter process that does not lower the gain of the high frequency component much more than the filter process in Step S103 as a decoding block. This is performed (S104). For example, the post filter 20 performs filter processing by selecting a coefficient that increases the cutoff frequency of the low-pass filter. Specifically, the post filter 20 uses a weighting factor of 1: 5: 1 for three consecutive pixels in the vertical direction in the vertical three-tap filter process. As a result, when the encoding accuracy is low, a detailed signal is stored by not reducing the gain of the high frequency component so much in the filter processing.

  As described above, the image decoding apparatus 100 according to Embodiment 1 of the present invention can perform adaptive filter processing according to the quantization parameter. Further, as described above, the post filter 20 performs a filtering process on the decoded image data in units of macroblocks, and therefore adaptively and simply performs a filtering process in accordance with the quantization parameter determined in units of macroblocks. Can be realized.

  Here, the decoded image data generated by the adding unit 4 and the local decoded image generated by the adding unit 57 of the image encoding device 600 need to match. Therefore, the loop-like process including the adder 4, the memory 5, and the motion compensation unit 6 needs to be the same as the loop-like process including the adder 57, the memory 58, and the motion compensation unit 59 of the image encoding device 600. is there. For example, in the conventional image decoding device 500 shown in FIG. 8, when the post filter 21 is arranged in the previous stage of the memory 5, the loop processing of the image decoding device 500 and the loop processing of the image encoding device 400 are performed. Will be different. That is, in the conventional image decoding apparatus 500, the post filter 21 cannot be arranged in the front stage of the memory 5 and the filter process cannot be performed in units of macro blocks.

  On the other hand, in the image decoding apparatus 100 according to Embodiment 1 of the present invention, the post-filter 20 is placed before the memory 15 that is necessary for decoding image data with 10-bit precision that is not included in the loop processing. By performing the processing, filter processing is performed in units of macroblocks. That is, the image decoding apparatus 100 according to Embodiment 1 of the present invention performs a loop-like process including the adder 4, the memory 5, and the motion compensator 6, as an adder 57, a memory 58, and an image encoder 600. Filter processing in units of macroblocks can be performed while maintaining the same processing as the loop processing composed of the motion compensation unit 59.

  As described above, the image decoding apparatus 100 according to Embodiment 1 of the present invention decodes encoded data using predicted image data with 8-bit accuracy, and outputs image data with 10-bit accuracy. That is, the loop-like process including the adder 4, the memory 5, and the motion compensation unit 6 of the image decoding device 100 is the loop-like process including the adder 57, the memory 58, and the motion compensation unit 59 of the image encoding device 600. The input signal, the processing content, and the bit accuracy are the same. Therefore, the local decoded image generated by the adding unit 57 matches the decoded image generated by the adding unit 4. Thereby, the image decoding apparatus 100 can decode the encoded data encoded by the image encoding apparatus 600. That is, the image decoding apparatus 100 decodes encoded data in which 10-bit precision image data is encoded in accordance with an image encoding standard that defines input / output of 8-bit precision image data. Can do.

  In addition, the loop-like process including the adder 4, the memory 5, and the motion compensation unit 6 of the image decoding device 100 is the loop-like process including the adder 57, the memory 58, and the motion compensation unit 59 of the image encoding device 400. The input signal, the processing content, and the bit accuracy are the same. Therefore, the local decoded image generated by the adding unit 57 matches the decoded image generated by the adding unit 4. Thereby, the image decoding apparatus 100 can decode the encoded data encoded by the image encoding apparatus 400. That is, the image decoding apparatus 100 decodes encoded data in which 8-bit precision image data is encoded in accordance with an image encoding standard that defines input / output of 8-bit precision image data. Can do.

  That is, the image decoding apparatus 100 according to Embodiment 1 of the present invention can decode a high-bit-accuracy image while maintaining compatibility with a normal encoding method (while maintaining standards compliance). it can. Therefore, the playback device including the image decoding device 100 according to Embodiment 1 of the present invention impairs compatibility with the conventional playback device even if the encoding method is already standardized broadcasting and recording media. Therefore, it is possible to transmit / receive or record / reproduce a good image with higher accuracy and less compression distortion.

  Furthermore, the post filter 20 of the image decoding apparatus 100 according to Embodiment 1 of the present invention performs a filtering process on the decoded image data in units of macroblocks before being stored in the memory 15. As a result, the image decoding apparatus 100 according to Embodiment 1 of the present invention has a circuit scale of the post filter 20 as compared with the image decoding apparatus 700 that performs the filtering process on the conventional decoded image in frame units or field units. Can be reduced. For example, when performing a filtering process in the vertical direction orthogonal to the horizontal direction in which pixel signals of pixels are sequentially read, the conventional image decoding device 700 requires a delay circuit for the number of pixels Nfh in the horizontal direction of the frame. On the other hand, the image decoding apparatus 100 according to Embodiment 1 of the present invention may include a delay circuit for the number of pixels Nmbh in the horizontal direction of the macroblock.

  Furthermore, the image decoding apparatus 100 according to Embodiment 1 of the present invention performs a filtering process on decoded image data input in units of macroblocks using the quantization parameter 30 determined in units of macroblocks. Thereby, the image decoding apparatus 100 according to the present invention can realize adaptive filter processing using a quantization parameter with simpler control.

  Furthermore, the post filter 20 of the image decoding apparatus 100 according to Embodiment 1 of the present invention performs a filtering process in which the gain of the high frequency component is greatly reduced when the quantization step is coarse and the encoding accuracy is low. Thereby, it is possible to reduce noise generated in the decoded image at the time of encoding with low accuracy. Further, the post filter 20 performs a filter process that does not significantly reduce the gain of the high frequency component when the quantization step is fine and the encoding accuracy is high. Thereby, highly accurate image data can be output as it is. That is, the image decoding apparatus 100 according to Embodiment 1 of the present invention can perform an optimal filter process according to the encoding accuracy.

  Furthermore, in the image decoding apparatus 100 according to Embodiment 1 of the present invention, the image data conversion unit 11 performs the 9-bit accuracy by performing a simple process such as rounding the lower bits of the 11-bit accuracy difference pixel value with high bit accuracy. Are formed. Therefore, the image decoding apparatus 100 according to Embodiment 1 of the present invention can easily generate differential pixel values with 9-bit accuracy and 11-bit accuracy.

  In the above description, the post filter 20 performs the 3-tap vertical filter process, but the number of taps is not limited to 3 taps. Further, the post filter 20 may perform horizontal filter processing. Further, the post filter 20 may use a filtering process including a filtering process between fields and a time direction between frames.

  In the above description, the post filter 20 performs the filtering process according to the quantization step, but the filtering process may be performed according to the picture type of the encoded data. Specifically, when the picture type is a B picture, the post filter 20 performs a filter process in which the gain of the high frequency component is greatly reduced. Specifically, the post filter 20 selects a coefficient that lowers the cutoff frequency of the low-pass filter and performs filter processing. For example, the post filter 20 uses a weighting factor of 1: 2: 1 for three consecutive pixels in the vertical direction in the vertical three-tap filter process. Further, when the picture type is an I picture or a P picture, the post filter 20 performs a filter process that does not significantly reduce the gain of the high frequency component. Specifically, the post filter 20 performs a filter process by selecting a coefficient that increases the cutoff frequency of the low-pass filter. For example, the post filter 20 uses a weighting factor of 1: 5: 1 for three consecutive pixels in the vertical direction in the vertical three-tap filter process. The I picture is a picture that performs predictive coding within a frame, the P picture is a picture that performs predictive coding between frames in one direction, and the B picture is a picture that performs prediction between frames in two directions. is there.

(Embodiment 2)
The image decoding apparatus according to Embodiment 2 of the present invention is a modification of the image decoding apparatus 100 according to Embodiment 1. The image decoding apparatus according to Embodiment 2 of the present invention is an example in which the configuration of the image decoding apparatus is changed without changing the feature of the present invention in which post-filter processing is added to decoded image data in units of macroblocks. is there.

  FIG. 4 is a block diagram showing the configuration of the image decoding apparatus according to Embodiment 2 of the present invention. Elements similar to those in FIG. 1 are denoted by the same reference numerals, and detailed description thereof is omitted.

  The image decoding apparatus 200 illustrated in FIG. 4 includes an inverse quantization inverse orthogonal transform unit 3 instead of the image data conversion unit 11 with respect to the image decoding apparatus 100 illustrated in FIG.

  The image decoding apparatus 200 is an image decoding apparatus that complies with an image encoding standard (for example, MPEG2) in which input / output of 8-bit precision image data is defined, and is obtained by encoding 8-bit precision image data. The encoded data is decoded. The image decoding apparatus 200 also decodes image data that has been variable-length encoded with 10-bit accuracy, and outputs 10-bit accuracy image data. For example, the image decoding apparatus 200 decodes the encoded data encoded by the image encoding apparatus 600 shown in FIG. 9 into 10-bit precision image data and outputs the decoded image data.

  The image decoding apparatus 200 includes a code input unit 1, a variable length decoding unit 2, an inverse quantization inverse orthogonal transform unit 3, an addition unit 4, a memory 5, a motion compensation unit 6, and an inverse quantum. The inverse inverse orthogonal transform unit 13, the addition unit 14, the memory 15, the post filter 20, and the image output unit 17 are provided.

  The inverse quantization inverse orthogonal transform unit 3 generates a difference pixel value with 9-bit accuracy by performing an inverse quantization process and an inverse orthogonal transform process on the quantized orthogonal transform coefficient generated by the variable length decoding unit 2. To do.

  The adder 4 adds the 9-bit accuracy difference pixel value generated by the inverse quantization inverse orthogonal transform unit 3 and the 8-bit accuracy predicted image data output from the motion compensation unit 6, thereby obtaining 8-bit accuracy. Decode image data in units of macro blocks is generated.

  With the above configuration, the image decoding apparatus 200 according to Embodiment 2 of the present invention maintains compatibility with a normal encoding scheme, as with the image decoding apparatus 100 according to Embodiment 1. High-bit precision images can be decoded. In addition, the image decoding apparatus 200 according to Embodiment 2 of the present invention has a circuit scale of the post filter 20 as compared with the conventional image decoding apparatus 700 that performs filter processing on image data in frame units or field units. Can be small. Also, the image decoding apparatus 200 according to Embodiment 2 of the present invention can realize adaptive filter processing using quantization parameters with simpler control.

(Embodiment 3)
The image decoding apparatus according to Embodiment 3 of the present invention is a modification of the image decoding apparatus 100 according to Embodiment 1. In Embodiment 3 of the present invention, an image coding standard using a deblocking filter (for example, the H.264 standard) is not changed without changing the feature of the present invention that post-filter processing is added to decoded image data in units of macroblocks. ) Will be described.

First, H. An image encoding device corresponding to the H.264 standard will be described.
FIG. 1 is a block diagram illustrating a configuration of an image encoding device corresponding to the H.264 standard. In addition, the same code | symbol is attached | subjected to the element similar to FIG. 9, and detailed description is abbreviate | omitted.

  An image encoding device 601 shown in FIG. 5 further includes a deblocking filter 71 in addition to the configuration of the image encoding device 600 shown in FIG.

  The deblocking filter 71 reduces noise generated at the macroblock boundary of the local decoded image generated by the adding unit 57. The filter characteristic of the deblocking filter 71 is determined by the image encoding device 601, and filter characteristic data indicating the characteristic is output from the image encoding device 601 together with the encoded image data.

  The memory 58 stores the local decoded image in which noise is reduced by the deblocking filter 71.

  With the above-described configuration, the image encoding device 601 has the H.264 format. It encodes 10-bit precision image data corresponding to the H.264 standard and outputs encoded data.

Next, an image decoding apparatus according to Embodiment 3 of the present invention will be described.
FIG. 6 is a block diagram showing a configuration of an image decoding apparatus according to Embodiment 3 of the present invention. Elements similar to those in FIG. 1 are denoted by the same reference numerals, and detailed description thereof is omitted.

  The image decoding apparatus 300 illustrated in FIG. 6 further includes a deblocking filter 31 in addition to the configuration of the image decoding apparatus 100 illustrated in FIG.

  The image decoding apparatus 300 shown in FIG. This is an image decoding device that complies with the H.264 standard and a profile that defines input / output of 8-bit precision image data. That is, the image decoding apparatus 300 receives 8-bit precision image data as H.264. The encoded data encoded by H.264 is decoded. The image decoding apparatus 300 decodes image data that has been variable-length encoded with 10-bit accuracy, and outputs 10-bit accuracy image data. For example, the image decoding apparatus 300 decodes the encoded data encoded by the image encoding apparatus 601 shown in FIG. 5 and outputs 10-bit precision image data.

  The image decoding apparatus 300 includes a code input unit 1, a variable length decoding unit 2, an inverse quantization inverse orthogonal transform unit 13, an image data transform unit 11, an adder 4, a memory 5, and a motion compensation unit. 6, an adder 14, a memory 15, a post filter 20, an image output unit 17, and a deblock filter 31.

  The deblock filter 31 reduces noise generated at the macroblock boundary of the decoded image data generated by the adder 4. At this time, the image decoding apparatus 300 determines the filter characteristic of the deblocking filter 31 using the filter characteristic data output from the image encoding apparatus 601. Therefore, the filter characteristic of the deblocking filter 71 in the image coding apparatus 601 matches the filter characteristic of the deblocking filter 31 in the image decoding apparatus 300.

  The memory 5 stores the decoded image data whose noise is reduced by the deblocking filter 31.

  Here, a loop-like process including the adder 4, the deblock filter 31, the memory 5, and the motion compensation unit 6 is performed from the adder 57, the deblock filter 71, the memory 58, and the motion compensation unit 59 of the image encoding device 601. The loop-shaped processing is the same as the input signal, processing content, and bit accuracy. As a result, the local decoded image generated by the adding unit 57 and the decoded image data generated by the adding unit 4 match.

  With the above configuration, the image decoding apparatus 300 according to Embodiment 3 of the present invention is an H.264 standard. Decoding corresponding to the H.264 standard can be performed.

  In addition, the image decoding apparatus 300 according to Embodiment 3 of the present invention maintains high compatibility with a normal encoding scheme and maintains a high bit rate, similarly to the image decoding apparatus 100 according to Embodiment 1. An accurate image can be decoded. Further, the image decoding apparatus 300 according to Embodiment 3 of the present invention has a circuit scale of the post filter 20 as compared with the conventional image decoding apparatus 700 that performs filter processing on image data in frame units or field units. Can be small. Also, the image decoding apparatus 300 according to Embodiment 3 of the present invention can realize adaptive filter processing using quantization parameters with simpler control.

  The present invention can be applied to an image decoding device, and in particular, can be applied as an image decoding device used for a digital television, a digital video camera, and a digital video recorder.

It is a block diagram which shows the structure of the image decoding apparatus which concerns on Embodiment 1 of this invention. It is a figure which shows the processing image of the filter process by the post filter of the image decoding apparatus concerning Embodiment 1 of this invention, and the post filter of the conventional image decoding apparatus. It is a flowchart which shows the flow of the filter process of the post filter of the image decoding apparatus which concerns on Embodiment 1 of this invention. It is a block diagram which shows the structure of the image decoding apparatus which concerns on Embodiment 2 of this invention. H. 1 is a block diagram illustrating a configuration of an image encoding device corresponding to the H.264 standard. It is a block diagram which shows the structure of the image decoding apparatus which concerns on Embodiment 3 of this invention. It is a block diagram which shows the structure of the conventional image coding apparatus. It is a block diagram which shows the structure of the conventional image decoding apparatus. It is a block diagram which shows the structure of the conventional image coding apparatus which is based on an image coding standard and performs high quality encoding. It is a block diagram which shows the structure of the conventional image decoding apparatus which conforms to an image coding standard and decodes with high image quality.

Explanation of symbols

100, 200, 300, 500, 700 Image decoding device 400, 600, 601 Image encoding device 1 Code input unit 2 Variable length decoding unit 3, 13, 56 Inverse quantization Inverse orthogonal transformation unit 4, 14, 57 Addition Unit 5, 15, 58 Memory 6, 59 Motion compensation unit 7, 17 Image output unit 11 Image data conversion unit 20, 21, 22 Post filter 30 Quantization parameter 31, 71 Deblock filter 51, 61 Image input unit 52, 62 Subtraction unit 53, 63 Orthogonal transform quantization unit 54 Variable length coding unit 55 Code output unit 201 Decoded image frame 211, 212, 213 Pixel 221, 222 Decoded image data in units of macroblocks

Claims (9)

Variable length decoding means for generating quantized orthogonal transform coefficients by variable length decoding encoded data;
By performing inverse quantization and inverse orthogonal transform on the quantized orthogonal transform coefficient, a first difference pixel value having a bit accuracy and a second difference pixel value having a higher accuracy than the first bit accuracy are generated. Inverse quantization inverse orthogonal transform means;
First memory means for holding one or more reference images of the first bit precision;
A reference image corresponding to the difference pixel value of the first bit accuracy generated by the inverse quantization inverse orthogonal transform unit among the one or more reference images held in the first memory unit; A first adding means for adding a value and causing the first memory means to hold as a reference image;
A reference image corresponding to the difference pixel value of the second bit precision generated by the inverse quantization inverse orthogonal transform unit among the one or more reference images held in the first memory unit; and the difference pixel Second addition means for generating image data in units of macroblocks of the second bit accuracy by adding a value;
Filter means for performing filter processing for reducing noise in the image data in units of macroblocks generated by the second addition means;
2nd memory means for storing the image data of the macroblock unit filtered by the filter means and outputting the image data of the second bit precision field unit or frame unit. Decryption device. The filter means, when dequantizing the quantized orthogonal transform coefficient corresponding to the image data of the macroblock unit, according to the quantization parameter used in the inverse quantization inverse orthogonal transform means, The image decoding apparatus according to claim 1, wherein a filtering characteristic applied to the image data of the block is changed. 3. The image decoding according to claim 2, wherein the filtering unit performs a filtering process with different filter characteristics corresponding to a size of a quantization step indicating a quantization interval included in the quantization parameter. apparatus. The inverse quantization inverse orthogonal transform means includes:
A first inverse quantization inverse orthogonal transform unit that generates the difference pixel value of the second bit precision by performing inverse quantization and inverse orthogonal transform on the quantized orthogonal transform coefficient;
Conversion means for converting the difference pixel value of the second bit precision generated by the first inverse quantization inverse orthogonal transform means into the difference pixel value of the first bit precision. The image decoding apparatus according to claim 1. The inverse quantization inverse orthogonal transform means includes:
A first inverse quantization inverse orthogonal transform means for generating the difference pixel value of the first bit precision by performing inverse quantization and inverse orthogonal transform on the quantized orthogonal transform coefficient;
And a second inverse quantization inverse orthogonal transform unit configured to generate the difference pixel value with the second bit precision by performing inverse quantization and inverse orthogonal transform on the quantized orthogonal transform coefficient. Item 2. The image decoding device according to Item 1. The image decoding device further includes:
A deblocking filter for reducing noise at a macroblock boundary of the image data added by the first adding means;
The image decoding apparatus according to claim 1, wherein the first memory unit holds image data whose noise is reduced by the deblocking filter as a reference image. A variable-length decoding step for generating a quantized orthogonal transform coefficient by variable-length decoding encoded data;
By performing inverse quantization and inverse orthogonal transform on the quantized orthogonal transform coefficient, a first difference pixel value having a bit accuracy and a second difference pixel value having a higher accuracy than the first bit accuracy are generated. An inverse quantization inverse orthogonal transform step;
A first accumulation step of holding one or more reference images of the first bit precision in a first memory means;
A reference image corresponding to the difference pixel value of the first bit precision generated in the inverse quantization inverse orthogonal transform step among one or more reference images held in the first memory means, and the difference pixel value A first addition step of causing the first memory means to hold the reference image as a reference image;
A reference image corresponding to the difference pixel value of the second bit precision generated in the inverse quantization inverse orthogonal transform step among the one or more reference images held in the first memory means; and the difference pixel A second addition step of generating image data in units of macroblocks of the second bit precision by adding values;
A filter step for performing filter processing for reducing noise on the image data in units of macroblocks generated in the second addition step;
A second accumulating step of accumulating the image data in units of macroblocks filtered in the filtering step in a second memory means;
An image decoding method comprising: outputting the second bit precision image data stored in the second memory means in a field unit or a frame unit. A variable-length decoding step for generating a quantized orthogonal transform coefficient by variable-length decoding encoded data;
By performing inverse quantization and inverse orthogonal transform on the quantized orthogonal transform coefficient, a first difference pixel value having a bit accuracy and a second difference pixel value having a higher accuracy than the first bit accuracy are generated. An inverse quantization inverse orthogonal transform step;
A first accumulation step of holding one or more reference images of the first bit precision in a first memory means;
A reference image corresponding to the difference pixel value of the first bit precision generated in the inverse quantization inverse orthogonal transform step among one or more reference images held in the first memory means, and the difference pixel value A first addition step of causing the first memory means to hold the reference image as a reference image;
A reference image corresponding to the difference pixel value of the second bit precision generated in the inverse quantization inverse orthogonal transform step among the one or more reference images held in the first memory means; and the difference pixel A second addition step of generating image data in units of macroblocks of the second bit precision by adding values;
A filter step for performing filter processing for reducing noise on the image data in units of macroblocks generated in the second addition step;
A second accumulating step of accumulating the image data in units of macroblocks filtered in the filtering step in a second memory means;
A program for causing a computer to execute an output step of outputting the second bit precision image data stored in the second memory means in a field unit or a frame unit. Variable length decoding means for generating quantized orthogonal transform coefficients by variable length decoding encoded data;
By performing inverse quantization and inverse orthogonal transform on the quantized orthogonal transform coefficient, a first difference pixel value having a bit accuracy and a second difference pixel value having a higher accuracy than the first bit accuracy are generated. Inverse quantization inverse orthogonal transform means;
First memory means for holding one or more reference images of the first bit precision;
A reference image corresponding to the difference pixel value of the first bit accuracy generated by the inverse quantization inverse orthogonal transform unit among the one or more reference images held in the first memory unit; A first adding means for adding a value and causing the first memory means to hold as a reference image;
A reference image corresponding to the difference pixel value of the second bit precision generated by the inverse quantization inverse orthogonal transform unit among the one or more reference images held in the first memory unit; and the difference pixel Second addition means for generating image data in units of macroblocks of the second bit accuracy by adding a value;
Filter means for performing filter processing for reducing noise in the image data in units of macroblocks generated by the second addition means;
And second memory means for storing the image data in units of macroblocks filtered by the filter means and outputting the image data in field units or frame units of the second bit precision. circuit.

Images