JP2018085604A - Loop filter, encoding device, decoding device, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a device and a program that obtain images with high quality by appropriately processing a hand-offset, i.e. a SAO (Sample Adaptive Offset) function of a loop filter even in image signals including OETF (opto-electric transformation) or color difference information different from before such as HDR signals and wide color range signals.SOLUTION: A loop filter 30 for performing a hand-offset process in a local decoding process or decoding process of an encoded video signal includes; a band division unit 31 for setting band division of the video signal to which the band-offset process is applied on the basis of evenness of an optical input signal; an offset parameter setting unit 32 for setting an object band to which an offset is applied; and an offset execution unit 33 for executing the offset process for the object band on the basis of a given offset value.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a loop filter, an encoding device, a decoding device, and a program, and in particular, a loop filter that performs offset processing in video compression encoding processing or decoding processing, and an encoding device, decoding device, and the like including the loop filter, and The present invention relates to a program for realizing these.

  In recent years, with the practical application of high-resolution video such as Super Hi-Vision, more realistic video has been demanded, and as a technical means for realizing a new expression of video, High Dynamic Range (HDR) is It is attracting attention (Non-Patent Document 1). This is a recommendation BT. Of ITU-R (International Telecommunication Union Radiocommunications Sector). In contrast to the standard dynamic range (SDR) defined in 709 (Non-Patent Document 2), the maximum luminance that can be displayed on the display is increased, so that the luminance range that can display the video signal is also expanded. As a result, it is possible to display a very bright luster that could not be expressed by SDR and a simultaneous high-quality display in the shade of the stadium.

  In order to realize HDR, a display capable of high luminance display is necessary as described above, and it is also necessary to support HDR signals for various cameras that perform photography and editing. Also, in applications such as broadcasting that distribute and transmit video, it is necessary to consider measures in video signal compression encoding processing. Although HDR has a wider range of brightness and color differences than SDR, the photoelectric conversion (OETF: Optical-Electronic Transfer Function) without changing the number of gradations (bit depth) per pixel that can be handled by conventional video equipment. It is the mainstream to deal with it by devising a non-linear conversion process (also called a gamma curve (conversion)). For this reason, in the video encoding process that handles a video signal as a waveform, the process is basically unchanged unless the number of gradations is changed. That is, video coding schemes such as MPEG (Moving Picture Experts Group) -2, AVC (Advanced Video Coding) /H.264, HEVC (High Efficiency Video Coding) /H.265 can be used.

  In addition to HDR, there is a wide color gamut as a technique for expanding the expression of video. This expands the color gamut that can be expressed by using RGB (red, green, blue), which are the three primary colors of light, as vivid colors as close to a single wavelength as possible. In conventional Hi-Vision, ITU-R recommendation BT. Although the three primary colors specified in 709 are used, it is recommended that the BT. The use of the three primary colors defined in 2020 will become the mainstream in the future. Regarding the wide color gamut, it is necessary for the video equipment to support, but in the video encoding process, if the bit depth corresponding to each color does not change, the conventional process can be used.

  In the latest video encoding method, HEVC / H.265 (Non-Patent Document 3), a loop filter is used. The loop filter is a filter that processes an image within a loop structure that performs local decoding processing in video encoding processing, but is not limited thereto, and is also used in decoding processing in a decoding device. The loop filter has a function called sample adaptive offset (SAO). This is a filter function that adds an offset for each pixel in order to correct distortion caused by quantization in the encoding process, and reduces DC component errors and mosquito noise caused by quantization.

  FIG. 8 shows an outline of image processing using sample adaptive offset. FIG. 9 shows specific details of the offset processing. There are two types of offsets: an edge offset (EO) shown in FIG. 8A and a band offset (BO) shown in FIG. 8B.

  Of these, the edge offset (EO) is to correct the pixel value of the processing target pixel in accordance with the relationship between the two pixels adjacent to the processing target pixel. Specifically, first, the pixel values of the processing target pixel (reconstruction pixel) and two adjacent pixels (pixels adjacent in the vertical, horizontal, and diagonal directions) are compared and classified into categories according to the magnitude relationship. When the relationship with the adjacent pixel is one of the four types of patterns shown in FIG. 9A, a category (edge index) is selected, and an offset value is specified and an offset process is performed. The edge offset can be corrected so that local peaks and valleys (ringing) of pixel values become flat as shown in FIG.

  On the other hand, the band offset (BO) divides a pixel value (luminance or color difference) equally into 32 bands based on the gradation, and applies an offset by a value determined for each band. As shown in FIG. 9B, an offset process is performed on four consecutive bands. The encoder side finds four consecutive bands that can reduce the square error energy the most, calculates the offset for each band, and gives the band position information (index that specifies the first band of the target band to perform the offset) and the offset value. Send to the decoder side. On the decoder side, the designated offset value is applied to all pixel values in the band. Since the band offset is processed according to the pixel value, it is effective when there is little variation in the pixel value within the processing range. In particular, this is effective when a low-frequency component quantization error or an inter-frame reference phase shift occurs due to motion compensation, and has an edge enhancement effect as shown in FIG. 8B.

ARIB STD-B67, “ESSENTIAL PARAMETER VALUES FOR THE EXTENDED IMAGE DYNAMIC RANGE TELEVISION (EIDRTV) SYSTEM FOR PROGRAM PRODUCTION”, Version 1.0, Association of Radio Industries and Businesses, July 3, 2015 Recommendation ITU-R BT.709, “Parameter values for the HDTV standards for production and international program exchange” Jun Okubo (supervised), “H.265 / HEVC Textbook”, Impress Japan Co., Ltd., October 21, 2013, p.157-161

  When the OETF (photoelectric conversion function) used for signal processing is different from the conventional SDR OETF, such as HDR, the video coding method itself can be used as it is, but depending on the video signal, the coding is performed. Deterioration may easily occur.

  FIG. 10 shows an example of the characteristics of OETF that converts an optical signal (optical input signal) into an electrical signal (video signal). The OETF is a non-linear curve graph having a shape similar to a logarithm or a power root curve with respect to an optical signal acquired by photographing a subject, and generally compresses a high luminance range to a predetermined bit depth. is there. The vertical axis of FIG. 10 is a value obtained by normalizing the intensity (gradation) of the video signal, and the horizontal axis indicates the intensity of the optical input signal. One of the optical input signals can be used as a reference white. If this curve has the shape of a logarithmic function according to human visual characteristics (Weber-Fechner law), the converted electrical signal is treated as a linear signal and processed without depending on the pixel value. If the display is finally performed after EOTF (Electronic-Optical Transfer Function), which is the reverse conversion of OETF, there is almost no problem in visual image quality. However, when the OETF is a function of a curve deviating from human visual characteristics (for example, when it has a dynamic range different from that of the conventional SDR method), the same processing is performed regardless of the pixel value by regarding the electric signal as linear. If applied, there is a risk of quality degradation depending on the pixel value.

  In the band offset processing defined by HEVC / H.265, the band is set by equally dividing the pixel value (the gradation of the video signal) as described above. FIG. 11 schematically shows band division set by dividing a video signal into equal bandwidths. When the pixel value is divided into 32, the video signal is equally divided into 32 pixel levels per band when the video signal is 10 bits (1024 pixel level) and 8 pixel levels per band when the video signal is 8 bits (256 pixel level). That is, as shown in FIG. 11, the band offset is performed by equally dividing the pixel value into equal parts to perform band division, and the specified band is offset with a constant value regardless of the pixel value. Therefore, when viewed as an optical signal, when the pixel value is high, the optical signal changes greatly due to the offset, but when the pixel value is low, the optical signal hardly changes even when the offset is applied. Therefore, it is expected that the intended offset does not match the actual image quality change.

  In addition, color signals are generally processed by converting into luminance and two color difference signals instead of RGB signals in video encoding processing. The purpose of this is to reduce the correlation between the luminance and the color difference, and to reduce (subsample) the resolution of the color difference signal based on the characteristic that human vision is insensitive to the color signal. In the color difference signal space in the luminance / color difference signal (for example, YCbCr), similarly to the luminance signal, the signal is not an optical signal but an electric signal, and the color difference signal space is not a uniform color space. Even if they are different, the way the difference appears depends on the color. For this reason, when band offset processing is performed, a change that is not an intended change in color may occur.

  Accordingly, an object of the present invention made in view of the above problems is to obtain a high-quality image by offset processing even for an image signal having characteristics different from those of a conventional signal such as an HDR signal and a wide color gamut signal. It is an object to provide a loop filter, an encoding device, a decoding device, and a program.

  In order to solve the above problems, a loop filter according to the present invention is a loop filter that performs band offset processing in local decoding processing or decoding processing of an encoded video signal, and the band of the video signal subjected to the band offset processing The division is set based on the uniformity of the optical input signal.

  In order to solve the above problems, a loop filter according to the present invention is a loop filter that performs band offset processing in local decoding processing or decoding processing of an encoded video signal, and the band of the video signal subjected to the band offset processing A band dividing unit that sets division based on the uniformity of the optical input signal, an offset parameter setting unit that sets a target band for performing offset, and a given offset value for the target band And an offset execution unit that executes offset processing.

  The loop filter performs band division of the video signal to be subjected to the band offset processing, and the intensity width of the optical input signal is equalized in each divided band based on the relationship between the optical input signal and the video signal. It is desirable to set as follows.

  The loop filter sets the band division of the video signal to be subjected to the band offset processing so that the color difference is equal in the uniform color space in each divided band based on the color signal space of the optical input signal. It is desirable to do.

  The loop filter may set the band division of the video signal to be subjected to the band offset processing so that the accumulated frequency is equal in each divided band based on the frequency distribution of the optical input signal. desirable.

  Further, it is desirable that the loop filter switches between band division that equally divides by a video signal and band division that equally divides by an optical input signal based on OETF information or color space information.

  In order to solve the above-described problem, an encoding apparatus according to the present invention is an encoding apparatus including a local decoding process of an encoded video signal, and performs a band offset process in the local decoding process using the loop filter. It is characterized by performing.

  In order to solve the above problems, a decoding apparatus according to the present invention performs a band offset process in the decoding process using the loop filter in a decoding apparatus that performs a decoding process of an encoded video signal. And

  In order to solve the above problems, a program according to the present invention causes a computer to function as the loop filter, the encoding device, or the decoding device.

  According to the loop filter, the encoding device, the decoding device, and the program of the present invention, video compression encoding is performed by adjusting the image quality by band offset even for an image signal having characteristics different from the conventional one such as an HDR signal and a wide color gamut signal. A high quality image can be obtained in the processing or decoding processing.

It is a block diagram which shows the example of the encoding apparatus of this invention. It is a block diagram which shows the example of a structure of the loop filter of this invention. It is a figure explaining the band division based on the intensity | strength of an optical input signal. It is a figure explaining the band division | segmentation in the uniform color space of an optical input signal. It is a figure explaining the band division based on a histogram. It is a flowchart which shows the example of operation | movement of the loop filter of this invention. It is a block diagram which shows the example of the decoding apparatus of this invention. An overview of image processing by sample adaptive offset is shown. The contents of the offset processing of the sample adaptive offset are shown. It is a graph which shows an example of the characteristic of OETF. It is a figure which shows typically the band division | segmentation which divided | segmented the video signal into equal band width.

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

(Embodiment 1)
A loop filter according to the present invention and an encoding apparatus including the loop filter will be described. FIG. 1 is a block diagram showing an example of an encoding apparatus according to Embodiment 1 of the present invention. In the description, the HEVC / H.265 system is assumed as the encoding / decoding process, but the loop filter of the present invention can also be applied to other encoding systems having processing similar to the band offset.

  The encoding apparatus 100 inputs an original image and outputs an encoded bit stream, and corresponds to OETF and color space processing when an original image (video signal) is generated from an optical input signal. The band offset processing of the loop filter 30 can be changed. Therefore, OETF information and / or color space information is input in addition to the original image.

  The encoding apparatus 100 includes a screen dividing unit 10, a subtractor 11, a transform / quantization unit 12, an entropy encoding unit 13, an inverse quantization / inverse transform unit 14, an adder 15, a deblocking filter 16, a loop filter 30, A motion detection unit 17, a motion compensation prediction unit 18, and an intra prediction unit 19 are provided. Among these, the inverse quantization / inverse transform unit 14, the adder 15, the deblocking filter 16, and the loop filter 30 function as a local decoding processing unit. The transform / quantization unit 12, the local decoding processing unit, the motion compensation prediction unit 18 or the intra prediction unit 19, and the subtractor 11 constitute a loop structure.

  The present invention differs from the prior art in the function and configuration of the loop filter 30. The loop filter of the present invention is characterized in that band division is performed based on the uniformity of the optical input signal in accordance with the OETF and color space of the input signal.

  The screen division unit 10 receives an original image and divides the screen into images of a predetermined size. The divided image data is output to the subtractor 11 and the motion detection unit 17, and the subsequent processing is performed for each image block.

  The subtractor 11 performs a subtraction process on the prediction image input from the motion compensation prediction unit 18 or the intra prediction unit 19 from the blocked image input from the screen division unit 10 to obtain a difference between both images, The difference image is output to the transform / quantization unit 12.

  The transform / quantization unit 12 performs, for example, encoding processing (discrete cosine transform, etc.) and quantization processing defined in HEVC / H.265 on the difference image input from the subtractor 11, The result is output to the entropy encoding unit 13 and the local decoding processing unit.

  The entropy encoding unit 13 entropy-encodes data necessary for decoding an image, such as OETF information / color space information, which will be described later, together with encoded image data input from the transform / quantization unit 12 (appearance for each symbol). (Encoding using different codeword lengths based on the probability) to create a bit stream, and then outputting to a transmission path and the like.

  The inverse quantization / inverse transform unit 14 performs inverse quantization and inverse transform (such as discrete cosine inverse transform) on the encoded data input from the transform / quantization unit 12. That is, a decoding process opposite to the process performed by the transform / quantization unit 12 is performed, and the result is output to the adder 15.

  The adder 15 is processed by the motion-compensated prediction unit 18 or the intra-prediction unit 19 to be described later, and the data that has been inverse-quantized and inverse-transformed by the inverse-quantization / inverse-conversion unit 14, that is, the data obtained by decoding the difference image The prediction image data thus added is added, and the combined image data is output to the deblocking filter 16 and the intra prediction unit 19.

  The deblocking filter 16 performs deblocking processing on the output image data from the adder 15. That is, in HEVC / H.265 and the like, processing in units of blocks is performed, so that discontinuous distortion (block distortion) may occur at the block boundary. The deblocking filter 16 performs block boundary filter processing to reduce block distortion. Then, the image data subjected to the deblocking process is output to the loop filter 30.

  The loop filter 30 performs the above-described sample adaptive offset processing on the image data subjected to the deblocking processing. The loop filter 30 of the present invention receives OETF information / color space information related to the generation of a video signal (original image), and performs a level of offset processing according to the OETF of the video signal and the color space in the band offset processing. Change the range banding method. Details of the configuration and operation of the loop filter 30 will be described later. The loop filter 30 outputs the filtered image data to the motion detection unit 17 and the motion compensation prediction unit 18.

  The motion detection unit 17 compares the image data input from the screen division unit 10 with the decoded image data input from the loop filter 30 of the local decoding processing unit, detects the motion of the image, and detects the detected motion The information (motion vector or the like) regarding is output to the motion compensation prediction unit 18.

  The motion compensation prediction unit 18 performs motion compensation prediction processing based on the image processed by the loop filter 30 such as band offset and the motion information of the image input from the motion detection unit 17 and subtracts the motion prediction image. To the adder 11 and the adder 15.

  The intra prediction unit 19 performs intra prediction, that is, intra prediction processing based on the output image of the adder 15, and outputs the intra prediction image to the subtractor 11 and the adder 15. Note that whether the motion prediction image from the motion compensation prediction unit 18 or the intra prediction image from the intra prediction unit 19 is output to the subtractor 11 can be appropriately selected by a switch.

  Next, the configuration and operation of the loop filter 30 will be described. In the HEVC / H.265 local decoding process of FIG. 1, a video signal that has been subjected to inverse quantization, inverse transform, motion compensation, and deblocking filter is input to the loop filter 30. At the same time, corresponding information is extracted from VUI (Video Usability Information) or SEI (Supplemental Enhancement Information) in which OETF information defining the OETF of the video signal and color space information defining the color space to be used are stored, and the loop filter 30 is extracted. Is input.

  FIG. 2 is a block diagram illustrating an example of the configuration of the loop filter 30. The loop filter 30 includes a band dividing unit 31, an offset parameter setting unit 32, and an offset execution unit 33. The loop filter 30 is given a predetermined offset mode (mode information specifying whether to perform band offset or edge offset). FIG. 2 shows the configuration of the loop filter 30 when performing band offset processing, and the configuration when performing edge offset is omitted.

  The band division unit 31 receives OETF information / color space information, and sets a band division that is a basis of a band offset based on this information. That is, the OETF information / color space information for generating the video signal is the same as the conventional SDR or BT. In the case of the conventional color gamut defined in 709, band division is performed by equal division with a video signal (electrical signal) as in the past, and OETF information / color space information for generating a video signal is HDR or BT. In the case of a wide color gamut defined by 2020, band division is performed based on the uniformity of the optical input signal, and each bandwidth is set. In addition, whether to perform the band division to divide equally with the conventional video signal or the band division to divide equally with the optical input signal, how much the optical-electrical conversion is with human visual characteristics (Weber-Fechner law) It is desirable to categorize whether they match (or diverge), and set an appropriate standard, compare the standard with the input OETF information / color space information, determine, and switch Just do it.

  In the case of a band offset, the offset parameter setting unit 32 receives information about an offset execution band (for example, an index for designating the first band of target bands to be offset) from a control unit (not shown), and is set by the band division unit 31. Based on the band division and the input information, a band (for example, four bands continuous from the designated band) to be offset is set.

  The offset execution unit 33 receives an offset value (offset value and sign) from a control unit (not shown), and performs band offset based on the designated offset value for the target band set by the offset parameter setting unit 32. Execute. The designated offset value is applied to all pixel values in the target band.

  Next, the band division setting process performed by the band division unit 31 will be described. In particular, three types of processing will be described in the case where the band dividing unit 31 performs band division based on the uniformity of the optical input signal based on the OETF information / color space information regarding the video signal. Note that the band division unit 31 has a switching function, and in the case of SDR, it is also possible to set band division that equally divides an electric signal (video signal) as usual.

<Response to HDR signal>
The band dividing unit 31 divides the video signal into a predetermined number of bands based on the input OETF information. Here, a case where the video signal and the OETF information are HDR signals will be described. For example, the number of bands can be divided into 32 according to the HEVC / H.265 system, but any number of bands can be used.

  In the OETF information, a relationship between an electrical signal (video signal) to be encoded and an optical signal (optical input signal) that is an image captured by a camera or the like is described by a mathematical expression, a look-up table (LUT), or the like. Yes. In the present invention, the band offset applied to the electrical signal by the HEVC / H.265 method or the like can be equivalently applied to the optical signal converted based on the OETF.

  FIG. 3 is a diagram for explaining band division based on the intensity of an optical signal (optical input signal). As an example, a level range (light intensity range) in an optical signal before conversion by OETF is divided into equal bandwidths (in the figure, divided into 8 equal parts, but may be divided into 32 equal parts, for example) The division position of the electric signal corresponding to the equal division of the optical signal is calculated from the function of the OETF, and the band division that is the basis of the band offset processing is set accordingly. Other processing is the same as HEVC / H.265 processing. Based on the offset execution band information, the specified offset value is offset for the band to which the specified offset is applied and up to three bands above that band. Apply processing. Note that the number of bands subjected to offset processing is not limited to four bands, and an optimal value can be appropriately set as the offset value. In the case of this example, since the offset is matched with the optical signal, this is equivalent to performing processing with a linear signal corresponding to linearity with actual light. Here, the OETF is described as being an HDR signal. However, even if the OETF other than HDR and deviating from human visual characteristics is subjected to band division processing in which the intensity width of the optical signal is similarly equal. Good.

<Support for wide color gamut signals>
Based on the input color space information, the band dividing unit 31 divides a color difference signal among video signals into a predetermined number of bands. Here, a case where the video signal and the color space information are wide color gamut signals will be described. In this case, based on the color space signal (color space information), the original YCbCr signal is converted into a uniform color space such as the Luv color space or the Lab color space via the RGB color space and the XYZ color space.

When the color space signal is ARIB STD-B56 (non-constant luminance transmission) that defines a wide color gamut signal, the conversion formula between RGB and YCbCr is:
Y '= 0.2627B' + 0.6780G '+ 0.0593R'
Cb '= (B'-Y') / 1.8814
Cr '= (R'-Y') / 1.4746
It has become. “′” In each variable indicates that a non-linear transfer function, so-called gamma correction is performed, and this gamma correction is the same as the above-described OETF.

  The RGB signal is subjected to inverse gamma correction and converted into an optical signal, and then converted into an XYZ color space. The conversion equation is shown as equation (1). Note that since the RGB color space is device-dependent, there are no single conversion formulas for XYZ.

  In the conversion of equation (1), for example, in the case of the sRGB (standard RGB) standard using D65 as the color temperature reference, the matrix M is expressed by the following equation (2).

Next, conversion to the L ^* , u ^* , v ^* space, which is a uniform color space, will be described. The conversion formula from the XYZ color space to the uv chromaticity diagram is
u '= 4X / (X + 15Y + 3Z)
v '= 9Y / (X + 15Y + 3Z)
It is expressed. Using these u ′ and v ′, the color coordinates (L ^* , u ^* , v ^* ) of the three-dimensional orthogonal coordinate space are defined by the following equation (3).

Y _n , u ′ _n , and v ′ _n are obtained from the white tristimulus values X, Y, and Z as a reference. Under D65 (daylight color), X _n = 95.04, Y _n = 100.00, and Z _n = 108.88.

Further, L ^* is a uniform color space, a ^*, b ^* As for conversion into the space, three-dimensional orthogonal coordinates from the XYZ color space (L ^*, a ^*, b ^*) conversion formula into a uniform color space of Is defined by the following equation (4).

  FIG. 4 shows band division in a uniform color space of an optical signal (optical input signal). An equal bandwidth is set in the converted uniform color space, and a reverse bandwidth is applied to the signal of the set bandwidth to obtain the bandwidth in the original YCbCr. That is, the band division is set based on the color signal space of the optical input signal so that the color difference is equal in the uniform color space in each divided band. By doing this, offset processing is performed after setting a bandwidth that matches the uniform color, and this is equivalent to performing processing with a signal corresponding to the actually perceived color.

<Band division based on histogram>
As an aspect different from the above processing, a method of performing SAO processing efficiently by automatically setting bandwidth when OETF information / color space information is not given will be described.

  FIG. 5 is a diagram for explaining band division based on a histogram. In the local decoding or decoding process, a histogram of the video signal YCbCr that has been subjected to inverse quantization, inverse transform, motion compensation, and deblocking filter is created. FIG. 5A is a histogram with the pixel value (luminance or color difference signal) of the video signal as the horizontal axis (level) and the vertical axis as the occurrence frequency. Since the video signal and the optical signal (optical input signal) have a one-to-one correspondence, the histogram of the video signal in FIG. 5 is equivalent to the frequency distribution of the optical input signal. The distribution of the signal is analyzed from this histogram, and the division band position of the luminance or color difference signal level is determined in accordance with the distribution.

  A cumulative histogram is created from the histogram of FIG. 5A, as shown in FIG. 5B, and is divided into a predetermined number of bands so that the vertical axis (cumulative frequency) has an equal width based on the maximum value of the cumulative distribution. . The pixel value (horizontal axis) is divided at a position where the cumulative frequency becomes equal, and the set bandwidth in the electric signal is determined. By such a procedure, division according to distribution can be automatically performed.

  By applying a band offset with the bandwidth determined in this way, an offset that is not wasteful can be applied in accordance with the video signal and the optical input signal.

  Next, the operation of the loop filter will be described. FIG. 6 is a flowchart showing an example of the operation executed by the loop filter.

  In step 1 (S1), it is determined whether or not OETF information / color space information is input. If there is an input, the process proceeds to Step 2, and if OETF information / color space information is not given, the process proceeds to Step 9.

  In step 2 (S2), it is determined whether the input OETF information / color space information is OETF information or color space information. If the information is OETF, that is, information relating to luminance, the process proceeds to step 3, that is, a processing step based on electrical / optical signal conversion. If the information is related to color difference signals, the process proceeds to step S6, that is, uniform color space conversion.

  In step 3 (S3), OETF which is a conversion function of electrical / optical signal conversion is read to determine what band division is performed, and the process proceeds to step 4.

  In step 4 (S4), based on the read OETF, for example, in the case of an HDR signal, band division is performed so that the optical signal has an equal bandwidth, each bandwidth in the video signal is set, and the process proceeds to step 5 .

  In step 5 (S5), offset processing is performed according to the offset execution band and the offset value based on the set bandwidth. The band offset related to the luminance signal ends here.

  Next, when it is determined in step 2 that the color difference signal (color space information), in step 6 (S6), it is determined what band division is to be performed. Based on the signal (YCbCr), conversion to a uniform color space (for example, Luv space or Lab space) is performed, and the process proceeds to step 7.

  In step 7 (S7), in the uniform color space processed in step 6, band division is performed so that the optical signal has an equal bandwidth, and the inverse of the color signal is performed to set each bandwidth in YCbCr. , Go to Step 8.

  In step 8 (S8), based on the set bandwidth, offset processing is performed according to the offset execution band and the offset value. This completes the band offset for the color difference signal.

  Next, when it is determined in step 1 that there is no OETF information / color space information, a histogram of the video signal YCbCr is created in step 9 (S9). Further, a cumulative histogram is created from the histogram, and the process proceeds to step 10.

  In step 10 (S10), the pixel value (horizontal axis) is divided at a position where the vertical axis (cumulative frequency) is equally divided based on the cumulative histogram, the set bandwidth in the video signal is determined, and the process proceeds to step 11 .

  In step 11 (S11), based on the set bandwidth, offset processing is performed according to the offset execution band and the offset value. The band offset based on the histogram ends here. Based on the above flowchart, band offset by a loop filter is performed.

  In the flowchart of FIG. 6, it has been described that the band division method of the loop filter is switched based on OETF information / color space information. However, a dedicated loop filter that performs any specific band division may be used. Is possible.

(Embodiment 2)
A decoding device will be described as a second embodiment. The loop filter is used not only in the local decoding process in the encoding apparatus but also in the decoding process of the decoding apparatus.

  FIG. 7 is a block diagram showing an example of a decoding device using the loop filter of the present invention. The decoding apparatus 200 receives a bit stream that is encoded data, performs a decoding process, and outputs a restored image.

  The decoding apparatus 200 includes an entropy decoding unit 20, an inverse quantization / inverse conversion unit 21, an adder 22, a deblocking filter 23, a loop filter 30, an intra prediction unit 24, and a motion compensation prediction unit 25. ing.

  The entropy decoding unit 20 performs entropy decoding on the input bit stream, and decodes data necessary for decoding an image such as OETF information / color space information together with the encoded data of the transmitted image. The encoded image data is output to the inverse quantization / inverse transform unit 21, and the OETF information / color space information is output to the loop filter 30.

  The inverse quantization / inverse transform unit 21 performs inverse quantization and inverse transform (such as discrete cosine inverse transform) on the encoded data input from the entropy decoding unit 20. That is, a decoding process opposite to the process performed by the encoding apparatus 100 is performed, and the result is output to the adder 22.

  The adder 22 includes data that has been inverse-quantized and inverse-transformed by the inverse-quantization / inverse-transformer 21 (this corresponds to difference image data), and a motion compensation predictor 25 or an intra-predictor 24 described later. The processed predicted image data is added, and the combined image data is output to the deblocking filter 23 and the intra prediction unit 24.

  The deblocking filter 23 performs deblocking processing on the output image data from the adder 22. That is, the filter processing of the block boundary part is performed to reduce discontinuous distortion (block distortion) generated at the block boundary. Then, the image data subjected to the deblocking process is output to the loop filter 30.

  The loop filter 30 performs a sample adaptive offset process on the image data input from the deblocking filter 23, particularly a band offset process. The loop filter 30 of the decoding device 200 has the configuration shown in FIG. 2, and sets the bandwidth of the band offset based on the OETF information / color space information input from the entropy decoding unit 20 and performs an offset process. . As already described, the loop filter of the present invention performs band division based on the uniformity of the optical input signal according to the OETF of the input signal and the color space. In other words, band division in which the intensity of the optical signal is equally divided (intensity width is uniform) or band division in which color differences are equal in a uniform color space is performed. Alternatively, the bandwidth may be set based on a histogram of the image signal. The loop filter 30 outputs the filtered image data to the motion compensation prediction unit 25 and also outputs it as an output image (restored image) of the decoding device 200.

  The intra prediction unit 24 performs intra prediction, that is, intra prediction processing based on the output image of the adder 22, and outputs the intra prediction image to the adder 22.

  The motion compensation prediction unit 25 performs a motion compensation prediction process based on an image that has been subjected to processing such as band offset by the loop filter 30 and motion prediction information of an image (not shown), and outputs a motion prediction image to the adder 22. .

  Note that whether the motion prediction image from the motion compensation prediction unit 25 or the intra-screen prediction image from the intra prediction unit 24 is output to the adder 22 can be appropriately selected by a switch.

  It should be noted that a computer can be suitably used to function as the loop filter 30, the encoding device 100, and the decoding device 200 described above, and such a computer can be used as the loop filter 30, the encoding device 100, and the decoding device 200. This can be realized by storing a program describing the processing contents for realizing each function in the storage unit of the computer, and reading and executing the program by a central processing unit (CPU) of the computer. This program can be recorded on a computer-readable recording medium.

  In the present embodiment, the configuration and operation of the encoding device 100 and the decoding device 200 including the loop filter 30 have been described. However, the present invention is not limited to this, and the band offset processing method in the loop filter and the encoding device 100 are described. May be configured as a coding method according to, and a decoding method according to the decoding apparatus 200.

  Although the above embodiment has been described as a representative example, it will be apparent to those skilled in the art that many changes and substitutions can be made within the spirit and scope of the invention. Therefore, the present invention should not be construed as being limited by the above-described embodiments, and various modifications and changes can be made without departing from the scope of the claims. For example, the functions included in each block, step, etc. can be rearranged so as not to be logically contradictory, and a plurality of constituent blocks, steps, etc. can be combined into one or divided. is there.

DESCRIPTION OF SYMBOLS 10 Screen division part 11 Subtractor 12 Transformation / quantization part 13 Entropy encoding part 14 Quantization / inverse conversion part 15 Adder 16 Deblocking filter 17 Motion detection part 18 Motion compensation prediction part 19 Intra prediction part 20 Entropy decoding part 21 Inverse quantization / inverse transform unit 22 Adder 23 Deblocking filter 24 Intra prediction unit 25 Motion compensation prediction unit 30 Loop filter 31 Band division unit 32 Offset parameter setting unit 33 Offset execution unit 100 Encoding device 200 Decoding device

Claims (9)

A loop filter that performs band offset processing in local decoding processing or decoding processing of an encoded video signal,
A loop filter characterized in that band division of a video signal to be subjected to the band offset processing is set based on uniformity of an optical input signal. A loop filter that performs band offset processing in local decoding processing or decoding processing of an encoded video signal,
A band division unit for setting the band division of the video signal to be subjected to the band offset processing based on the uniformity of the optical input signal;
An offset parameter setting unit for setting a target band for executing the offset;
An offset execution unit that performs an offset process on the target band based on a given offset value;
A loop filter comprising: The loop filter according to claim 1 or 2,
The band division of the video signal to be subjected to the band offset processing is set based on the relationship between the optical input signal and the video signal so that the intensity width of the optical input signal is uniform in each of the divided bands. A loop filter. The loop filter according to claim 1 or 2,
A loop characterized in that band division of a video signal to be subjected to band offset processing is set based on a color signal space of an optical input signal so that color differences are equal in a uniform color space in each divided band filter. The loop filter according to claim 1 or 2,
A loop filter characterized in that band division of a video signal to be subjected to the band offset processing is set based on an occurrence frequency distribution of an optical input signal so that cumulative frequency is equal in each divided band. A loop filter according to any one of claims 1 to 5,
A loop filter characterized by switching between band division equally divided by a video signal and band division equally divided by an optical input signal based on OETF information or color space information. In an encoding device including local decoding processing of an encoded video signal,
The encoding apparatus characterized by performing the band offset process in the said local decoding process using the loop filter as described in any one of Claims 1 thru | or 6. In a decoding device that performs decoding processing of an encoded video signal,
The decoding apparatus characterized by performing the band offset process in the said decoding process using the loop filter as described in any one of Claims 1 thru | or 6.   A program for causing a computer to function as the loop filter according to any one of claims 1 to 6, the encoding device according to claim 7, or the decoding device according to claim 8.

Images