JP6597385B2 - Encoding device and program, decoding device and program, and distribution system - Google Patents

Description

  The present invention relates to an encoding device and program, a decoding device and program, and a distribution system. For example, the present invention can be applied to a system that efficiently compresses an image divided into an important region and an unimportant region.

  In recent years, surveillance cameras have become widespread, and further higher frame rates, higher resolutions, and multi-viewpoints are desired. However, increasing the frame rate, increasing the resolution, and increasing the number of viewpoints cause an increase in the amount of moving image data, resulting in an increase in communication costs and storage costs. In order to alleviate this problem, conventionally, a method has been proposed in which a face area is detected from a moving image of a human image and a large number of bits are allocated to the compression of the face area.

  For example, Patent Document 1 proposes to reduce the amount of information by filtering using a low-pass filter in addition to the region of interest and removing high-frequency components. In the image encoding system described in Patent Document 1, filtering processing is performed on an image to be encoded in a region other than the region of interest using a prefilter unit, and the amount of information other than the region of interest is reduced, and then the compression unit Perform compression processing.

  In Patent Document 1, the prefilter unit receives the attention area information and the input image, and performs low-pass filtering only on the non-attention area in the input image, thereby outputting an information amount reduced image. It has a pass filter section. In the technique described in Patent Document 1, the amount of information is reduced by filtering with a low-pass filter and dropping high-frequency components. The technique described in Patent Document 1 uses the low-pass filter to produce an effect of reducing the amount of information and suppressing artifacts that occur at block boundaries.

  The reduction of the high frequency component by the low-pass filter corresponds to an operation for reducing the amount of information by converting the entire image into the frequency domain and reducing the amplitude of the high frequency component, for example.

  However, in many conventional compression techniques, each block is converted to the frequency domain and compressed. For example, in the conventional compression technique, even when the amplitude of the high frequency component is made zero by the low-pass filter, the amplitude of the high frequency component is not surely zero in the frequency region of the block unit.

  Therefore, the conventional compression technology can suppress the generation of artifacts, but unlike the case where the quality of the entire image is regarded as important as in the entertainment image, the background artifact is large as in the surveillance camera. For non-problematic applications, it is required to reduce the amount of data more efficiently and reliably.

  The technique described in Patent Document 2 uses a form in which the amount of information in a non-attention area is reduced by controlling parameters given to the compression unit. That is, the technique described in Patent Document 2 corresponds to a function of reducing the amount of information in the non-attention area according to the supplied parameter in the compression unit. As described above, in the technique described in Patent Document 2, it is possible to efficiently and surely reduce the data amount by reducing the information amount using the function provided by the compression unit.

Japanese Patent Laid-Open No. 04-219089 JP 2009-49979 A

  The technique described in Patent Document 2 is based on the premise that, as described above, the compression unit supports the function of reducing the information amount of the non-attention area according to the supplied parameter.

  However, when realizing image coding using the technology described in Patent Document 2, the compression unit is actually equipped with the function even though it has a function for controlling quality for each area according to the standard. There may be no cases. In particular, when the compression unit is provided as hardware, it is difficult to add the function. Even in such a case, a technique such as Patent Document 1 capable of controlling the video quality for each region without depending on the compression unit is desired.

  By the way, when inverse quantization is performed after quantization in the conventional prefilter unit, the statistical distribution of data (pixel information and coefficient information) input to the entropy encoder in the compression unit is a discrete distribution (that is, , Quantized distribution). On the other hand, H. H.264 and H.264. The distribution of information sources assumed by entropy encoders of many encoding systems such as H.265 is a continuous distribution such as a Laplace distribution or a Gaussian distribution. In the conventional entropy encoder, there is a problem that the compression efficiency is lowered when the statistical distribution of input data is different from the assumed distribution.

  In a conventional entropy encoder, one of the solutions is to optimize the distribution of the assumed information source in accordance with the prefilter unit. However, for example, in the case of a standardized encoding method, the distribution of information sources assumed by the entropy encoder may not be optimized to the distribution generated by the prefilter.

  In addition, in a conventional entropy encoder, an encoding method that can flexibly optimize the distribution of information sources assumed by the entropy encoder in accordance with the prefilter unit is also conceivable. However, even in this case, in the conventional entropy encoder, if the distribution is matched to the characteristics of the non-target block that is quantized and de-quantized by the pre-filter unit, the target block that is not quantized and de-quantized by the pre-filter unit Therefore, the compression efficiency is lowered in encoding the block of interest. In order to solve this problem, in the conventional entropy encoder, it is necessary to switch the distribution assumed by the region. Therefore, the encoding scheme employed adopts the distribution assumed by the entropy encoder according to the region. It is necessary to have a mechanism for switching.

  Even if the conventional entropy encoder has the above-described mechanism, the distribution assumed by the entropy encoder needs to be shared with the decoder. Overhead occurs.

  In view of the problems as described above, an encoding apparatus, a program, a decoding apparatus, and a decoding apparatus that can efficiently reduce the amount of information of an image without causing a problem of a decrease in compression efficiency and a narrowing of encoding method options. Programs and distribution systems are desired.

  According to a first aspect of the present invention, in an encoding device for encoding an input image, (1) attention block information describing whether each block constituting the input image is a target block or a non-target block; A pre-filter unit that reduces an information amount of the input image using a quantization parameter used for quantization of a block that constitutes the input image and an inverse quantization parameter, and generates an information amount reduced image; (2 An encoding unit that encodes the information amount reduced image; and (3) an output unit that outputs output data including encoded data encoded by the encoding unit, and (4) the prefilter unit (4-1) a conversion unit that converts the input image into blocks and outputs a conversion block; and (4-2) a non-target region based on the target block information among the conversion blocks output by the conversion unit. A transform block indicating that there is a quantization block that is quantized according to the quantization parameter and is output as a quantization block, and (4-3) the quantization unit outputs Of the quantized blocks, the quantized block indicating that it is a non-focused area in the focused block information is replaced with a block that has been subjected to inverse quantization according to the inverse quantization parameter, and transformed after reducing the amount of information. An inverse quantization unit that outputs as a block; and (4-4) an inverse transform unit that outputs, as the information amount reduced image, an image obtained by inversely transforming the transformed block sequence after the information amount reduction outputted by the inverse quantization unit. It is characterized by that.

  The decoding apparatus according to the second aspect of the present invention provides (1) attention block information describing whether each block constituting the input image is a target block or a non-target block, a quantization parameter, and an inverse quantization parameter. And a decoding unit that decodes encoded data obtained by encoding an information amount reduced image that has been subjected to a prefiltering process that reduces the information amount of the input image, and acquires the information amount reduced image; 2) an inverse quantization correction parameter acquisition unit that acquires an inverse quantization correction parameter based on the quantization parameter and the inverse quantization parameter used in the pre-filter processing; A post-filter unit that obtains a decoded image by filtering based on block information and information including the inverse quantization correction parameter, and (4) the post-filter unit (4-1) a conversion unit that converts the information-amount-reduced image into blocks and outputs the first conversion block sequence; and (4-2) a first conversion block sequence that the conversion unit outputs. The first transform block indicating the non-target region in the target block information is replaced with a block subjected to inverse quantization according to the inverse quantization correction parameter, and output as a second transform block sequence An inverse quantization correction unit that performs (4-3) an inverse transformation unit that inversely transforms the second transform block sequence output by the inverse quantization correction unit to obtain a decoded image and outputs the decoded image; It is characterized by having.

  According to a third aspect of the present invention, there is provided an encoding program that allows a computer installed in an encoding device that encodes an input image to: (1) whether each block constituting the input image is a target block or a non-target block; The amount of information of the input image is reduced by using the block information of interest described, the quantization parameter used for quantization of the block constituting the input image, and the inverse quantization parameter, and the information amount reduced image A prefilter unit to generate, (2) an encoding unit that encodes the information amount reduced image, and (3) an output unit that outputs output data including encoded data encoded by the encoding unit (4) the prefilter unit (4-1) converts the input image into blocks and outputs a converted block; and (4-2) a conversion block output by the converter. Among them, the transform block indicating the non-target region in the target block information is replaced with a block subjected to quantization according to the quantization parameter, and a quantization unit that outputs as a quantized block; (4-3) Out of the quantization blocks output by the quantization unit, inverse quantization according to the inverse quantization parameter is performed on the quantization block indicating the non-target region in the target block information. An inverse quantization unit that replaces the performed block and outputs the transformed block after the information amount reduction; and (4-4) an image obtained by inversely transforming the transformed block block after the information amount output by the inverse quantization unit. And an inverse conversion unit that outputs the reduced amount image.

  A decoding program according to a fourth aspect of the present invention provides: (1) attention block information describing whether a block is a target block or a non-target block for each block constituting the input image; Using the quantization parameter and the inverse quantization parameter, the encoded data obtained by encoding the information reduced image subjected to the prefiltering process for reducing the information amount of the input image is decoded to reduce the information amount. A decoding unit that acquires an image; (2) an inverse quantization correction parameter acquisition unit that acquires an inverse quantization correction parameter based on the quantization parameter and the inverse quantization parameter used in the prefiltering; and (3) Post-fill for obtaining a decoded image by filtering the information amount reduced image based on the information including the block-of-interest information and the inverse quantization correction parameter (4) the post filter unit (4-1) converts the information amount reduced image into blocks, and outputs a first converted block sequence; (4-2) Of the first transform block sequence output from the transform unit, the first transform block indicating that the target block information indicates a non-target region is subjected to inverse quantization according to the inverse quantization correction parameter. (4-3) the inverse transform of the second transform block sequence output from the inverse quantization corrector and the decoded image by inverse transform And an inverse transform unit that obtains and outputs the decoded image.

  A fifth aspect of the present invention is a distribution provided with an encoding device that encodes an input image to generate encoded data, and a decoding device that decodes the encoded data generated by the encoding device to generate a decoded image. In the system, the encoding apparatus of the first aspect of the present invention is applied as the encoding apparatus, and the decoding apparatus of the second aspect of the present invention is applied as the decoding apparatus.

  According to the present invention, it is possible to efficiently reduce the amount of information of an image without causing problems such as a decrease in compression efficiency and a narrowing of encoding method options.

It is the block diagram shown about the functional structure of the image coding system which concerns on 1st Embodiment. It is the block diagram shown about the functional structure of the image decoding system which concerns on 1st Embodiment. It is the block diagram shown about the whole structure of the delivery system which concerns on 1st Embodiment. It is explanatory drawing shown about the attention block used in 1st Embodiment. It is the flowchart shown about operation | movement of the image coding system which concerns on 1st Embodiment. It is the flowchart shown about operation | movement of the image decoding system which concerns on 1st Embodiment. It is the block diagram shown about the functional structure of the image coding system which concerns on 2nd Embodiment. It is the block diagram shown about the functional structure of the image decoding system which concerns on 2nd Embodiment. It is the flowchart shown about operation | movement of the image coding system which concerns on 2nd Embodiment. It is the flowchart shown about operation | movement of the image decoding system which concerns on 2nd Embodiment.

(A) First Embodiment Hereinafter, a first embodiment of an encoding device and program, a decoding device and program, and a distribution system according to the present invention will be described in detail with reference to the drawings. Hereinafter, an example in which the encoding apparatus of the present invention is applied to an image encoding system and the decoding apparatus of the present invention is applied to an image decoding system will be described.

(A-1) Configuration of the First Embodiment FIG. 3 is a block diagram showing the overall configuration of the distribution system 1 according to the first embodiment.

  The distribution system 1 encodes an input image and outputs a stream (bit string), and an image that decodes the stream (bit stream) encoded by the image encoding system 1001 and outputs a decoded image A decoding system 1501.

  The medium for inputting the stream output from the image encoding system 1001 to the image decoding system 1501 is not limited. For example, the stream output from the image encoding system 1001 may be transmitted to the image decoding system 1501 by communication (for example, the Internet), or the data of the stream output from the image encoding system 1001 may be data. It may be recorded on a recording medium (for example, a medium such as a DVD or HDD) and input to the image decoding system 1501 offline.

  In this embodiment, the image coding system 1001 is described as performing coding for each input image. However, the image coding system 1001 can process a plurality of input images in succession to perform moving image coding processing. You may make it apply. Similarly, the image decoding system 1501 may be applied to moving image decoding processing by successively decoding a plurality of encoded data streams.

  Next, the internal configuration of the image encoding system 1001 will be described. FIG. 1 is a block diagram showing a functional configuration of an image encoding system 1001 according to the first embodiment.

  The image encoding system 1001 includes an asymmetric quantization type prefilter unit 1002 and a compression unit 1009.

  The asymmetric quantization type prefilter unit 1002 reduces the information amount of the input image based on the block information of interest, the quantization parameter, and the inverse quantization parameter, and information on the information amount reduced image and the parameter used for the information amount reduction. A certain transmission quantization parameter is output.

  The target block information is information that determines whether or not the target region is a block unit.

  As a method of calculating the target block information, for example, as shown in FIG. 4, if even one pixel is included in the conversion block, it is determined that the conversion block is the target block, or the conversion block If more than half of the number of pixels in the attention area is included in the attention area, it is determined that the conversion block is the attention block, or if all the pixels in the conversion block are included in the attention area, There is a method of determining that the conversion block is a target block. Note that FIG. 4 shows an example in which a block of interest is determined when more than half of the pixels are included in the region of interest. In the image coding system 1001, the method for holding the target block information is not limited. For example, for each input image, the input of the corresponding target block information may be received from the outside. The target block information may be generated by analysis.

  In addition, the method of holding the quantization parameter and the inverse quantization parameter in the asymmetric quantization type prefilter unit 1002 is not limited, and the parameter set by the user or the like may be held. You may make it receive the input of the parameter corresponding for every image.

  The compression unit 1009 converts the information amount reduced image into, for example, JPEG or H.264. H.264, H.C. The stream is compressed (encoded) by a compression method (encoding method) such as H.265, the quantization parameter for transmission is also multiplexed, and a stream is output. That is, the compression unit 1009 is applied with a compression method (encoding method) using various entropy encoders.

  The asymmetric quantization type prefilter unit 1002 includes a transform unit 1003, a quantization unit 1005, an inverse quantization unit 1006, an inverse transform unit 1007, and a transmission quantization parameter generation unit 1008.

  The conversion unit 1003 converts the input image for each block and outputs a converted block. The transform unit 1003 receives an input image, performs, for example, two-dimensional orthogonal transform represented by discrete cosine transform for each block of an arbitrary size, and outputs a transform block.

  The quantization unit 1005 replaces only the block indicated as the non-target region in the target block information among the transform blocks supplied from the transform unit 1003 with a block quantized in accordance with the quantization parameter, thereby obtaining a “quantization block”. Is output as The quantization unit 1005 performs quantization according to the quantization parameter on the transform block indicated by the block-of-interest information as a non-target region among the transform blocks supplied from the transform unit 1003, Output as. In addition, the quantization unit 1005 outputs a transform block, which is indicated as the attention area by the attention block information, as a quantization block without being processed.

  The inverse quantization unit 1006 performs the inverse only for the block indicated as the non-target region by the target block information (that is, the block quantized by the quantization unit 1005) among the quantized blocks supplied from the quantizer 1005. It is replaced with a block that has been inversely quantized according to the quantization parameter, and is output as a “transform block after reducing the amount of information”. Further, when the block information of interest indicates that the region of interest is the region of interest, the inverse quantization unit 1006 outputs the transformed block after reducing the information amount without processing.

  The inverse transformation unit 1007 performs inverse transformation on the transformed block after information amount reduction and outputs an information amount reduced image.

  The transmission quantization parameter generation unit 1008 generates a transmission quantization parameter from the quantization parameter and the inverse quantization parameter. Details of the quantization parameter for transmission will be described later.

  Next, the internal configuration of the image decoding system 1501 will be described. FIG. 2 is a block diagram illustrating a functional configuration of the image decoding system 1501 according to the first embodiment.

  The image decoding system 1501 includes a decoding unit 1502 and a post filter unit 1503.

  The decoding unit 1502 decodes the stream, outputs an information amount reduced image, and has a function of separating the transmission quantization parameter from the stream.

  The post filter unit 1503 generates a decoded image based on the target block information, the information amount reduced image, and the transmission quantization parameter.

  The post filter unit 1503 includes an inverse quantization correction parameter generation unit 1504, a conversion unit 1505, an inverse quantization correction unit 1506, and an inverse conversion unit 1507.

  The inverse quantization correction parameter generation unit 1504 generates an inverse quantization correction parameter based on the transmission quantization parameter separated from the stream. Details of the transmission quantization parameter and the inverse quantization correction parameter will be described later.

  The conversion unit 1505 converts the information amount reduced image and generates a converted block after the information amount reduction.

  The inverse quantization correction unit 1506 generates a transform block by dequantizing the transformed block after reducing the amount of information based on the inverse quantization correction parameter when the noticed block information indicates a non-target region. It is.

  The inverse transform unit 1507 performs inverse transform on the transform block and generates a decoded image.

(A-2) Operation of the First Embodiment Next, the operation of the distribution system 1 (the image encoding system 1001 and the image decoding system 1501) of the first embodiment having the above configuration will be described.

  FIG. 5 is a flowchart showing the operation of the image encoding system 1001.

  First, the conversion unit 1003 divides the input image into blocks, converts them, and outputs them as converted blocks (S1001). A conversion method performed by the conversion unit 1003 includes, for example, discrete cosine conversion.

  The quantization unit 1005 performs quantization according to the quantization parameter for a block indicating that the target block information is a non-target region among the transform blocks, and outputs the quantized block (S1002). As the quantization parameter, for example, a quantization step width at the time of quantization or a right shift amount may be applied. When the quantization parameter is the quantization step width, the quantization unit 1005 obtains the coefficient of the quantization block, for example, by dividing the coefficient of the transform block by the quantization step width. For example, when the coefficient of the transform block is 100 and the quantization parameter is the quantization step width 20, the coefficient of the quantization block is 100 ÷ 20 = 5.

  The inverse quantization unit 1006 performs inverse quantization according to the inverse quantization parameter on the block indicating that the target block information is a non-target region among the quantized blocks, and outputs the result as a transform block after reducing the information amount (S1003).

  As the inverse quantization parameter, for example, a quantization step width and a left shift amount when performing inverse quantization can be applied. When the inverse quantization parameter is the quantization step width, the inverse quantization unit 1006 obtains the coefficient of the transformed block after the information amount reduction by multiplying the coefficient of the quantization block by the quantization step width, for example. For example, when the coefficient of the quantization block is 5 and the inverse quantization parameter is the quantization step width 2, the coefficient of the transform block after reducing the information amount is 5 × 2 = 10. Even if the inverse quantization parameter is a value at which the quantization step width becomes zero, the effect of the present invention can be obtained.

  The inverse transform unit 1007 inversely transforms the transformed block after the information amount reduction to generate an information amount reduced image (S1004).

  The inverse transformation process performed by the inverse transformation unit 1007 is an inverse transformation corresponding to the transformation performed in step S1001.

  Next, the transmission quantization parameter generation unit 1008 generates a transmission quantization parameter from the quantization parameter and the inverse quantization parameter (S1005). Details regarding the quantization parameter for transmission will be described later.

  Next, the compression unit 1009 converts the information amount reduced image into a JPEG or H.264 format. H.264, H.C. The stream is compressed by a compression method such as H.265, the transmission quantization parameter is multiplexed, and a stream is output (S1006).

  FIG. 6 is a flowchart showing the operation of the image decoding system 1501.

  First, the decoding unit 1502 decodes a stream, outputs an information amount reduced image, and separates a transmission quantization parameter from the stream (S1501).

  Next, the inverse quantization correction parameter generation unit 1504 generates an inverse quantization correction parameter from the separated transmission quantization parameter (S1502).

  Next, the conversion unit 1505 converts the information amount reduced image, and generates (restores) a converted block after the information amount reduction (S1503). The conversion performed by the conversion unit 1505 is the same as the conversion performed in step S1001.

  Next, the inverse quantization correction unit 1506 performs inverse quantization according to the inverse quantization correction parameter for a block indicating that the target block information is a non-target region among the transformed blocks after the information amount reduction. The converted block is output (S1504). As the inverse quantization correction parameter, for example, a quantization step width at the time of inverse quantization or a left shift amount can be applied.

  When the inverse quantization correction parameter is the quantization step width, the inverse quantization correction unit 1506 obtains the coefficient of the transform block after reducing the information amount by, for example, multiplying the coefficient of the quantization block by the quantization step width. . For example, when the coefficient of the transform block after the information amount reduction is 10 and the inverse quantization correction parameter is the quantization step width 10, the transform block coefficient is 10 × 10 = 100.

  Next, the inverse transform unit 1507 inversely transforms the transform block to generate a decoded image (S1505). The inverse transformation performed by the inverse transformation unit 1507 is an inverse transformation corresponding to the transformation performed in step S1503.

  Next, details of the inverse quantization correction parameter will be described.

  The inverse quantization correction parameter is a value obtained from the quantization parameter and the inverse quantization parameter. Quantization is performed by performing inverse quantization based on the inverse quantization parameter and inverse quantization based on the inverse quantization correction parameter. It is only necessary that the existence range of data narrowed by quantization based on the quantization parameter satisfies the return of the original existence range.

  Therefore, as the inverse quantization correction parameter, for example, a quantization step width when performing inverse quantization or a left shift amount can be applied.

  When the quantization parameter, the inverse quantization parameter, and the inverse quantization correction parameter are, for example, a quantization step width, the inverse quantization correction parameter can be obtained by dividing the quantization parameter by the inverse quantization parameter. it can. Further, when the quantization parameter, the inverse quantization parameter, and the inverse quantization correction parameter are, for example, left and right shift amounts, the size of the inverse quantization correction parameter is the same as the size of the quantization parameter. It is obtained by subtracting the size of the quantization parameter.

  As described above, data composed of a quantization parameter and an inverse quantization parameter can be applied to the transmission quantization parameter. For example, the transmission quantization parameter may be composed of a combination of a plurality of data, or may be composed of an inverse quantization correction parameter calculated from the quantization parameter and the inverse quantization parameter.

(A-3) Effects of First Embodiment According to the first embodiment, the following effects can be achieved.

  In the first embodiment, the image decoding system 1501 handles a part of the inverse quantization process, so that the statistical distribution of data input to the compression unit 1009 is a continuous distribution (that is, a distribution with a small quantization step). ) And approaches the distribution assumed by the compression unit 1009 (entropy encoder) (that is, a continuous distribution such as a Laplace distribution and a Gaussian distribution), so that the reduction in compression efficiency of the compression unit 1009 (entropy encoder) is alleviated. The

(B) Second Embodiment Hereinafter, a second embodiment of the encoding device and program, the decoding device and program, and the distribution system according to the present invention will be described in detail with reference to the drawings. Hereinafter, an example in which the encoding apparatus of the present invention is applied to an image encoding system and the decoding apparatus of the present invention is applied to an image decoding system will be described.

(B-1) Configuration of Second Embodiment The overall configuration of a distribution system 1A according to the second embodiment can also be shown using FIG. Hereinafter, differences of the second embodiment from the first embodiment will be described.

  The distribution system 1A is different from the first embodiment in that an image encoding system 1001 and an image decoding system 1501 are replaced with an image encoding system 2001 and an image decoding system 2501.

  FIG. 7 is a block diagram showing a functional configuration of the image coding system 2001. In FIG. 7, the same or corresponding parts as those in FIG.

  The image coding system 2001 includes an asymmetric quantization type prefilter unit 2002 and a compression unit 1009.

  The compression unit 1009 performs the same processing as in the first embodiment, and thus detailed description thereof is omitted.

  The asymmetric quantization type prefilter unit 2002 includes a transform unit 1003, a DC component filtering unit 2004, a quantization unit 1005, an inverse quantization unit 2006, an inverse transform unit 1007, and a transmission quantization parameter generation unit 2008. .

  In the second embodiment, the transform unit 1003, the quantization unit 1005, and the inverse transform unit 1007 have the same configuration as in the first embodiment, and thus detailed description thereof is omitted.

  The DC component filtering unit 2004 smoothes the DC component of the transform block and outputs a transform block after reducing the DC information amount.

  The quantization unit 1005 quantizes the transformed block after reducing the DC information amount and outputs the quantized block.

  The inverse quantization unit 2006, based on the block-of-interest information, the DC inverse quantization parameter, and the AC inverse quantization parameter, applies a DC component inverse quantization parameter (hereinafter referred to as “DC inverse quantization parameter”) to the DC component. And inverse quantization based on the inverse quantization parameter for the AC component (hereinafter referred to as “AC inverse quantization parameter”).

  Note that, as described above, the inverse quantization correction parameter generation unit 2504 (inverse quantization correction unit 2506) uses the DC inverse quantization parameter and the AC inverse quantization parameter as the inverse quantization parameter. In the inverse quantization correction parameter generation unit 2504 (inverse quantization correction unit 2506), the method for holding the DC inverse quantization parameter and the AC inverse quantization parameter is not limited, and the parameter set by the user or the like is held. Alternatively, input of a parameter corresponding to each input image may be accepted (for example, accepted by information multiplexed in a stream).

  As described above, the inverse quantization unit 2006 performs frequency separation type inverse quantization processing.

  The transmission quantization parameter generation unit 2008 generates a transmission quantization parameter from the quantization parameter, the DC inverse quantization parameter, and the AC inverse quantization parameter. The configuration of the transmission quantization parameter in the second embodiment will be described later.

  FIG. 8 is a block diagram showing a functional configuration of the image decoding system 2501. In FIG. 8, the same parts or corresponding parts as those in FIG.

  The image decoding system 2501 includes a decoding unit 1502 and a post filter unit 2503.

  In the second embodiment, the decoding unit 1502 performs the same processing as in the first embodiment, and thus detailed description thereof is omitted.

  The post filter unit 2503 includes an inverse quantization correction parameter generation unit 2504, a conversion unit 1505, an inverse quantization correction unit 2506, and an inverse conversion unit 1507.

  In the second embodiment, the conversion unit 1505 and the inverse conversion unit 1507 have the same configuration as that of the first embodiment, and thus detailed description thereof is omitted.

  The inverse quantization correction parameter generation unit 2504 generates a DC inverse quantization correction parameter and an AC inverse quantization correction parameter from the transmission quantization parameter. Detailed configurations of the DC inverse quantization correction parameter and the AC inverse quantization correction parameter will be described later.

  The inverse quantization correction unit 2506 performs inverse quantization based on the DC inverse quantization correction parameter for the DC component, and performs inverse quantization based on the AC inverse quantization correction parameter for the AC component. As described above, the inverse quantization correction unit 2506 performs frequency separation type inverse quantization correction.

(B-2) Operation of Second Embodiment Next, the operation of the distribution system 1A (image encoding system 2001 and image decoding system 2501) of the first embodiment having the above-described configuration will be described.

  FIG. 9 is a flowchart showing the operation of the image coding system 2001.

  First, the conversion unit 1003 divides an input image into blocks and converts them, and outputs a converted block (S2001). A conversion method performed by the conversion unit 1003 includes, for example, discrete cosine conversion.

  Next, the DC component filtering unit 2004 smoothes the DC component of the transform block, and outputs a transform block after reducing the DC information amount (S2002). The DC component filtering unit 2004 smoothes the DC component by using the DC component of the adjacent converted block after information amount reduction. For example, the DC component filtering unit 2004 adds the DC components of adjacent transform blocks after reducing the amount of information, and smoothes the result by dividing the sum by the number of added DC components.

  Next, the quantization unit 1005 performs quantization according to the quantization parameter on the block indicating that the target block information is the non-target region among the transformed blocks after the DC information amount reduction, and the quantization block (S2003).

  Next, the inverse quantization unit 2006 performs inverse quantization according to the DC inverse quantization parameter or the AC inverse quantization parameter for the block indicating that the target block information is a non-target region among the quantized blocks. And output as a converted block after reducing the amount of information (S2004). Specifically, the inverse quantization unit 2006 performs inverse quantization on the DC component based on the DC inverse quantization parameter, and performs inverse quantization on the AC component based on the AC inverse quantization parameter. The inverse quantization unit 2006 sets the quantization step width to a value equal to the quantization step width of the quantization parameter in the DC inverse quantization parameter, and sets the AC inverse quantization parameter to the quantization step width of zero. Even if the value is set, the effect of the present invention can be obtained.

  Next, the inverse transformation unit 1007 inversely transforms the transformed block after the information amount reduction to generate an information amount reduced image (S2005). The inverse transformation performed by the inverse transformation unit 1007 is an inverse transformation corresponding to the transformation performed in step S2001.

  Next, the transmission quantization parameter generation unit 2008 generates a transmission quantization parameter from the quantization parameter, the DC inverse quantization parameter, and the AC inverse quantization parameter (S2006).

  In the second embodiment, data composed of a quantization parameter, a DC inverse quantization parameter, and an AC inverse quantization parameter can be applied as the transmission quantization parameter. In the second embodiment, the transmission quantization parameter may be composed of a combination of a plurality of data, or the DC inverse quantum calculated from the quantization parameter, the DC inverse quantization parameter, and the AC inverse quantization parameter. Alternatively, the correction correction parameter and the AC inverse quantization correction parameter may be used.

  Next, the compression unit 1009 converts the information amount reduced image into a JPEG or H.264 format. H.264, H.C. The stream is compressed by a compression method such as H.265, the transmission quantization parameter is multiplexed, and a stream is output (S2007).

  FIG. 10 is a flowchart showing the operation of the image decoding system 2501.

  First, the decoding unit 1502 decodes a stream, outputs an information amount reduced image, and separates a transmission quantization parameter from the stream (S2501).

  Next, the inverse quantization correction parameter generation unit 2504 generates a DC inverse quantization correction parameter and an AC inverse quantization correction parameter from the separated transmission quantization parameters (S2502).

  The DC inverse quantization correction parameter can be generated by the same processing as the inverse quantization correction parameter of the first embodiment using the quantization parameter and the DC inverse quantization parameter. The AC inverse quantization correction parameter can be generated by the same processing as the inverse quantization correction parameter of the first embodiment using the quantization parameter and the AC inverse quantization parameter.

  Next, the conversion unit 1505 converts the information amount reduced image, and generates a converted block after information amount reduction (S2503). The conversion performed by the conversion unit 1505 is the same as the conversion performed in step S2001.

  Next, the inverse quantization correction unit 2506 performs inverse quantization according to the DC inverse quantization correction parameter or the AC inverse quantization correction parameter when the target block information indicates a non-target region ( S2504).

  Specifically, the inverse quantization correction unit 2506 performs inverse quantization based on the DC inverse quantization correction parameter for the DC component, and inverse quantization based on the AC inverse quantization correction parameter for the AC component. I do.

  Next, the inverse transform unit 1507 inversely transforms the transform block to generate a decoded image (S2505). The inverse transformation is an inverse transformation corresponding to the transformation performed in step S2503.

(B-3) Effects of Second Embodiment According to the second embodiment, the following effects can be achieved.

  In the second embodiment, the DC component inverse quantization parameter and the AC component inverse quantization parameter are separated. Specifically, in the second embodiment, for example, by increasing the quantization step width represented by the DC inverse quantization parameter to the same level as the quantization step width represented by the quantization parameter, the post filter according to the present invention. Even if an image decoding system (for example, an image decoding system based only on a standard) that does not include a section is prepared, a non-attention area is an image that can be visually identified.

  In addition, depending on the encoding method employed in the compression unit and decoding unit, there is a method that calculates and compresses the difference between the attention area and the non-attention area, and in this case, a large difference between the attention area and the non-attention area. May occur, and the coding efficiency may be adversely affected. However, even when such a coding method is employed, in the second embodiment, the difference in DC component between the attention area and the non-attention area becomes small. For this reason, the code amount after compression becomes small.

  Furthermore, in the second embodiment, compared to the first embodiment, in the image coding system 2001, the DC component is subjected to inverse quantization with a large quantization step width. Therefore, the compression unit 1009 (entropy code) The compression efficiency in the generator is reduced, but the DC component filtering unit 2004 can reduce the amount of information of the DC component, so that a reduction in compression efficiency and the like can be suppressed.

(C) Other Embodiments The present invention is not limited to the above-described embodiments, and may include modified embodiments as exemplified below.

  (C-1) In each of the above-described embodiments, information related to quantization is shared between the image encoding system and the image decoding system through multiplexing of the transmission quantization parameters. May be shared in other ways. For example, the inverse quantization correction parameter is a value obtained from the quantization parameter and the inverse quantization parameter, and by performing inverse quantization based on the inverse quantization parameter and inverse quantization based on the inverse quantization correction parameter, The same effect can be obtained if the existence range of the data narrowed by the quantization based on the quantization parameter satisfies the property that the original existence range returns.

  (C-2) In the second embodiment, the AC inverse quantization parameter and the AC inverse quantization correction parameter are described as one value, but are a set of one or more values, and the frequency Even if the separation-type inverse quantization unit 2006 and the frequency separation-type inverse quantization correction unit 2506 distribute one or more values for each band and perform inverse quantization based on the distributed values. There is an effect. In this case, the transmission quantization parameter is composed of the quantization parameter and one or more inverse quantization parameters, or is calculated from the quantization parameter and one or more inverse quantization parameters. One or more inverse quantization correction parameters.

  (C-3) In the second embodiment, the quantization parameter is described as one value, but is a set of one or more values, and the quantization unit 1005 includes one or more values. Values may be assigned to each band, and quantization may be performed based on the assigned values. At this time, the frequency separation type inverse quantization unit 2006 and the frequency separation type inverse quantization correction unit 2506 distribute one or more values for each band and perform inverse quantization based on the distributed values. Can achieve the same effect. In this case, the transmission quantization parameter is composed of one or more quantization parameters and one or more inverse quantization parameters, or one or more quantization parameters and one or more quantization parameters. One or more inverse quantization correction parameters calculated from the inverse quantization parameters are included. In this case, the same effect can be obtained even if the value assigned to the DC component among the quantization parameters is a value at which the quantization step width becomes zero. In this case, even if an image decoding system (for example, an image decoding system based only on a standard) that does not include the post filter unit of the present invention is prepared, the non-attention area is an image that can be sufficiently visually identified.

  (C-4) In each of the above embodiments, the quantization parameter, the inverse quantization parameter, and the inverse quantization correction parameter are described as parameters used only in the non-target region. However, in each of the above embodiments, the quantization parameter is a set of the quantization parameter for the attention area and the quantization parameter for the non- attention area. In each of the above embodiments, the inverse quantization parameter is a set of the inverse quantization parameter for the attention area and the inverse quantization parameter for the non- attention area. Further, in each of the above embodiments, the inverse quantization correction parameter is a set of the inverse quantization correction parameter for the attention area and the inverse quantization correction parameter for the non- attention area.

  Therefore, in each of the above embodiments, the quantization unit of the image coding system quantizes the attention area based on the attention area quantization parameter, and the non- attention area based on the non- attention area quantization parameter. You may make it quantize. Further, in each of the above embodiments, the inverse quantization unit of the image coding system performs inverse quantization on the attention area based on the inverse quantization parameter for the attention area, and the non- attention area inverse quantization on the non- attention area. Inverse quantization may be performed based on the quantization parameter. Further, in each of the above embodiments, the inverse quantization correction unit of the image decoding system performs inverse quantization on the attention area based on the inverse quantization correction parameter for the attention area, and reverses the non- attention area for the non- attention area. Inverse quantization may be performed based on the quantization correction parameter. In each of the above embodiments, as described above, the same effect can be obtained even if the quantization parameter, the inverse quantization parameter, and the inverse quantization correction parameter are used.

  From the same concept, in each of the above embodiments, the same effect can be obtained even if the quantization parameter, the inverse quantization parameter, and the inverse quantization correction parameter are parameters that are used only in the region of interest. it can.

  (C-5) In the second embodiment, the DC component filtering unit 2004 may be excluded and the transform block may be directly input to the quantization unit 1005.

  DESCRIPTION OF SYMBOLS 1 ... Distribution system, 1001 ... Image coding system, 1002 ... Asymmetric quantization type pre-filter part, 1003 ... Transformation part, 1005 ... Quantization part, 1006 ... Inverse quantization part, 1007 ... Inverse transformation part, 1008 ... For transmission Quantization parameter generation unit, 1009 ... compression unit, 1501 ... image decoding system, 1502 ... decoding unit, 1503 ... post filter unit, 1504 ... inverse quantization correction parameter generation unit, 1505 ... conversion unit, 1506 ... inverse quantization correction unit , 1507 ... Inverse conversion unit.

Claims (18)

In an encoding device for encoding an input image,
Target block information describing whether the block is a target block or a non-target block for each block constituting the input image, a quantization parameter used for quantization of the block constituting the input image, and an inverse quantization parameter And a pre-filter unit that reduces an information amount of the input image and generates an information amount reduced image;
An encoding unit for encoding the information amount reduced image;
An output unit that outputs output data including encoded data encoded by the encoding unit;
The pre-filter unit is
A conversion unit that converts the input image into blocks and outputs a converted block;
Of the transform blocks output by the transform unit, transform blocks indicating non-target regions in the target block information are replaced with blocks that have been quantized according to the quantization parameter, and quantized blocks A quantization unit that outputs as
Of the quantized blocks output by the quantizing unit, the quantized block indicating that it is a non-target region in the target block information is replaced with a block subjected to inverse quantization according to the inverse quantization parameter. An inverse quantization unit that outputs a transformed block after reducing the amount of information;
An encoding device, comprising: an inverse transform unit that outputs an image obtained by inversely transforming the transformed block sequence after information amount reduction output from the inverse quantization unit as the information amount reduced image. Further comprising: a transmission quantization parameter generating unit that generates and outputs a transmission quantization parameter that is information related to each parameter used for information amount reduction in the information amount reduction image and the prefilter unit;
The encoding apparatus according to claim 1, wherein the output unit outputs a stream of data obtained by multiplexing the encoded data encoded by the encoding unit and the transmission quantization parameter. .   The encoding apparatus according to claim 2, wherein the transmission quantization parameter generated by the transmission quantization parameter generation unit includes the quantization parameter and the inverse quantization parameter. .   The transmission quantization parameter generated by the transmission quantization parameter generation unit includes an inverse used when inverse quantization related to the information amount reduced image is performed based on the quantization parameter and the inverse quantization parameter. The encoding apparatus according to claim 2, wherein a quantization correction parameter is included.   The inverse quantization correction parameter is a value obtained from the quantization parameter and the inverse quantization parameter, and for the information amount reduced image, inverse quantization based on the inverse quantization parameter and the inverse quantization are performed. The data existence range narrowed by the quantization based on the quantization parameter by performing inverse quantization based on the quantization correction parameter is a parameter capable of returning the original existence range. 5. The encoding device according to 4. The quantization parameter is a quantization step width when the quantization unit quantizes,
The inverse quantization parameter is a quantization step width when the inverse quantization unit performs inverse quantization.
The encoding apparatus according to claim 4 or 5, wherein the inverse quantization correction parameter is obtained by dividing the quantization parameter by the inverse quantization parameter. The quantization parameter is a right shift amount when the quantization unit quantizes,
The inverse quantization parameter is a left shift amount when the inverse quantization unit performs inverse quantization.
The encoding according to claim 4 or 5, wherein the inverse quantization correction parameter is a left shift amount obtained by subtracting the magnitude of the quantization parameter by the magnitude of the inverse quantization parameter. apparatus. The inverse quantization parameter includes a DC inverse quantization parameter and an AC inverse quantization parameter,
The inverse quantization unit performs inverse quantization on the DC component based on the DC inverse quantization parameter, and performs inverse quantization on the AC component based on the AC inverse quantization parameter. The encoding device according to any one of claims 1 to 7. Reduces the amount of information in the input image by using block-of-interest information that describes whether each block is a block of interest or non-block of interest, quantization parameters, and inverse quantization parameters. A decoding unit that decodes encoded data obtained by encoding an information amount reduced image that has been subjected to pre-filter processing, and acquires the information amount reduced image;
An inverse quantization correction parameter acquisition unit that acquires an inverse quantization correction parameter based on the quantization parameter and the inverse quantization parameter used in the prefiltering;
A post-filter unit that obtains a decoded image by filtering the information amount reduced image based on the information including the block-of-interest information and the inverse quantization correction parameter;
The post filter section is
A conversion unit that converts the information-amount-reduced image into blocks, and outputs the converted image as a first converted block sequence;
Of the first transform block sequence output by the transform unit, the first transform block indicating that the target block information is a non-target region is subjected to inverse quantization according to the inverse quantization correction parameter. An inverse quantization correction unit that outputs the second transformed block sequence in place of the performed block;
A decoding apparatus comprising: an inverse transform unit that inversely transforms the second transform block sequence output by the inverse quantization correction unit to obtain a decoded image and outputs the decoded image.   The inverse quantization correction parameter is a value obtained from the quantization parameter and the inverse quantization parameter, and for the information amount reduced image, inverse quantization based on the inverse quantization parameter and the inverse quantization are performed. The data existence range narrowed by the quantization based on the quantization parameter by performing inverse quantization based on the quantization correction parameter is a parameter capable of returning the original existence range. 9. The decoding device according to 9.   The decoding device according to claim 10, wherein the decoding unit acquires and decodes the encoded data from a stream including the encoded data. In the stream, transmission quantization parameters including the quantization parameter and the inverse quantization parameter are multiplexed,
The decoding apparatus according to claim 11, wherein the inverse quantization correction parameter acquisition unit acquires the inverse quantization correction parameter based on the transmission quantization parameter. The quantization parameter is a quantization step width used when performing quantization in the prefiltering process,
The inverse quantization parameter is a quantization step width used when inverse quantization is performed in the prefiltering process,
The decoding device according to claim 10, wherein the inverse quantization correction parameter is obtained by dividing the quantization parameter by the inverse quantization parameter. The quantization parameter is a right shift amount when quantizing by the prefiltering process,
The inverse quantization parameter is a left shift amount when inverse quantization is performed in the prefiltering process,
The inverse quantization correction parameter is a left shift amount when the inverse quantization correction unit performs inverse quantization, and is obtained by subtracting the size of the quantization parameter by the size of the inverse quantization parameter. The decoding device according to claim 10, wherein the decoding device is characterized in that The inverse quantization correction parameter includes a DC inverse quantization correction parameter for inverse quantization of the DC component, and an AC inverse quantization correction parameter for inverse quantization of the AC component,
The inverse quantization correction unit performs inverse quantization based on the DC inverse quantization correction parameter for a DC component, and performs inverse quantization based on the AC inverse quantization correction parameter for an AC component. The decoding device according to any one of claims 9 to 14. A computer mounted on an encoding device that encodes an input image,
Target block information describing whether the block is a target block or a non-target block for each block constituting the input image, a quantization parameter used for quantization of the block constituting the input image, and an inverse quantization parameter And a pre-filter unit that reduces an information amount of the input image and generates an information amount reduced image;
An encoding unit for encoding the information amount reduced image;
Function as an output unit that outputs output data including encoded data encoded by the encoding unit;
The pre-filter unit is
A conversion unit that converts the input image into blocks and outputs a converted block;
Of the transform blocks output by the transform unit, transform blocks indicating non-target regions in the target block information are replaced with blocks that have been quantized according to the quantization parameter, and quantized blocks A quantization unit that outputs as
Of the quantized blocks output by the quantizing unit, the quantized block indicating that it is a non-target region in the target block information is replaced with a block subjected to inverse quantization according to the inverse quantization parameter. An inverse quantization unit that outputs a transformed block after reducing the amount of information;
An encoding program comprising: an inverse transform unit that outputs an image obtained by inversely transforming the transformed block sequence after information amount reduction output by the inverse quantization unit as the information amount reduced image. The computer installed in the decryption device
Reduces the amount of information in the input image by using block-of-interest information that describes whether each block is a block of interest or non-block of interest, quantization parameters, and inverse quantization parameters. A decoding unit that decodes encoded data obtained by encoding an information amount reduced image that has been subjected to pre-filter processing, and acquires the information amount reduced image;
An inverse quantization correction parameter acquisition unit that acquires an inverse quantization correction parameter based on the quantization parameter and the inverse quantization parameter used in the prefiltering;
The information amount reduced image is made to function as a post filter unit that obtains a decoded image by performing filter processing based on the information including the block-of-interest information and the inverse quantization correction parameter,
The post filter section is
A conversion unit that converts the information-amount-reduced image into blocks, and outputs the converted image as a first converted block sequence;
Of the first transform block sequence output by the transform unit, the first transform block indicating that the target block information indicates a non-target region is subjected to inverse quantization according to the inverse quantization correction parameter. An inverse quantization correction unit that outputs the second transformed block sequence in place of the performed block;
A decoding program, comprising: an inverse transform unit that inversely transforms the second transform block sequence output by the inverse quantization correction unit to obtain a decoded image and outputs the decoded image.   In the distribution system, comprising: an encoding device that encodes an input image to generate encoded data; and a decoding device that generates the decoded image by decoding the encoded data generated by the encoding device. A distribution system in which the encoding device according to claim 1 is applied, and the decoding device according to claim 10 is applied as the decoding device.

Images