CN113591878A - Dynamic HDR image feature extraction method - Google Patents

Dynamic HDR image feature extraction method Download PDF

Info

Publication number
CN113591878A
CN113591878A CN202110781871.1A CN202110781871A CN113591878A CN 113591878 A CN113591878 A CN 113591878A CN 202110781871 A CN202110781871 A CN 202110781871A CN 113591878 A CN113591878 A CN 113591878A
Authority
CN
China
Prior art keywords
pixel
group
component
calculating
rgb
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202110781871.1A
Other languages
Chinese (zh)
Other versions
CN113591878B (en
Inventor
石小明
许国忠
孙伟涛
张贝利
肖玉超
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hangzhou Arcvideo Technology Co ltd
Original Assignee
Hangzhou Arcvideo Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hangzhou Arcvideo Technology Co ltd filed Critical Hangzhou Arcvideo Technology Co ltd
Priority to CN202110781871.1A priority Critical patent/CN113591878B/en
Publication of CN113591878A publication Critical patent/CN113591878A/en
Application granted granted Critical
Publication of CN113591878B publication Critical patent/CN113591878B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/51—Indexing; Data structures therefor; Storage structures

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Software Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Color Television Systems (AREA)
  • Facsimile Image Signal Circuits (AREA)
  • Color Image Communication Systems (AREA)

Abstract

The embodiment of the invention discloses a dynamic HDR image feature extraction method, which comprises the following steps: inputting a YUV image, grouping pixel points of the input image, and recording all pixel points sharing a group of UV as a pixel group; traversing each pixel group, calculating R, G, B component values of all pixel points in each pixel group, sharing UV values during calculation, taking Y component values of corresponding positions from Y pixel values, and calculating RGB component values of other pixels in the group through one pixel; calculating the maximum value in the R, G and B components of a certain pixel in the pixel group, and recording the index of the maximum value; calculating histogram statistics of RGB components and maximum components in the pixel group according to the RGB component values and the recorded maximum component index; and performing traversal accumulation or merging on the histograms of each group of pixel groups, and calculating RGB of the whole image and RGB maximum component histogram statistics.

Description

Dynamic HDR image feature extraction method
Technical Field
The invention belongs to the technical field of audio and video coding and transcoding processing, and particularly relates to a dynamic HDR image feature extraction method.
Background
In a Dynamic HDR (High-Dynamic Range, High Dynamic Range image) encoding and transcoding processing system, Dynamic HDR metadata of each frame of video needs to be calculated and extracted in real time and embedded into a code stream. In the dynamic HDR metadata calculation process, the luminance distribution of the image needs to be counted as a basis for generating subsequent dynamic metadata parameters.
YUV (also known as YCrCb) is a color coding method adopted by european television systems. In a modern color television system, a three-tube color camera or a color CCD camera is usually used for image capture, then the obtained color image signals are subjected to color separation and respective amplification and correction to obtain RGB, a luminance signal Y and two color difference signals R-Y (i.e., U) and B-Y (i.e., V) are obtained through a matrix conversion circuit, and finally a transmitting end respectively encodes the luminance and two color difference signals, and transmits the encoded luminance and two color difference signals through the same channel. The color representation method is the YUV color space. The storage format of the YUV code stream is closely related to the sampling mode thereof, and the mainstream sampling mode has three types: (1) YUV 4: 4, (2) YUV 4: 2, and (3) YUV 4: 2: 0. According to fig. 1, the solid point is Y, and the circle is a group of UV, where each Y corresponds to a pixel in the image (each pixel is a point in the image, for example, the resolution is 1920 × 1080, which means 1920 × 1080 ═ 2073600 pixels). In the YUV 4: 4 sampling, every 1Y corresponds to 1 group of UV components, and 1 pixel corresponds to 1Y, 1U, 1V. In YUV 4: 2 sampling, every 2 Ys share 1 group of UV components, and 2 pixels correspond to 2 Ys, 1U and 1V. In the YUV 4: 2: 0 sampling, 1 group of UV components is shared by every 4 ys, and 4 pixels correspond to 4 ys, 1U, 1V.
In the RGB color space, a model that quantitatively expresses colors by using the luminances of three primary colors of red, green, and blue is also called an additive color mixing model, and a method of realizing color mixing by superimposing three colors of RGB light on each other is suitable for display by a luminous body such as a display. Each pixel in an image (a frame of video) is represented by R, G, B3 components, so a set of RGB corresponds to an image pixel.
A Histogram (Histogram), also called a quality distribution map, is a statistical report graph in which a series of vertical stripes or line segments with different heights represent the distribution of data. The data type is generally represented by the horizontal axis, and the distribution is represented by the vertical axis. For the image histogram, the horizontal axis represents the value of a certain color component, and the vertical axis represents the frequency (number) of the occurrence of the value, and if the RGB components are stored in a 10-bit quantization manner, the image value range of the certain color component of each pixel point is 0 to 1024, the histogram of the certain color component is the image value of 0 to 1024 on the horizontal axis, and the frequency (number) of the occurrence of each value of the component on the vertical axis.
In the prior art, a general processing flow in a dynamic HDR metadata calculation process is as follows:
1) YUV420 or YUV422 is converted to YUV 444.
2) YUV444 is converted to RGB.
3) The R, G, B maximum value of the component (i.e. the rgbMax value) of each pixel is calculated in the RGB color space, then histogram, luminance information statistics, etc. are performed according to the RGB and rgbMax 4 values of each pixel, and finally, dynamic metadata extraction is performed by using the image statistical information.
In the processing method, each stage is independently processed, and each pixel is independently processed without considering the association relationship of the pixel points.
Disclosure of Invention
Aiming at the problems, the invention is used for providing a dynamic HDR image feature extraction method, improving the overall processing capacity of encoding and transcoding equipment and realizing dynamic HDR encoding processing with higher resolution and frame rate.
In order to solve the technical problems, the invention adopts the following technical scheme:
a dynamic HDR image feature extraction method comprises the following steps:
inputting a YUV image, grouping pixel points of the input image, and recording all pixel points sharing a group of UV as a pixel group;
traversing each pixel group, calculating R, G, B component values of all pixel points in each pixel group, sharing UV values during calculation, taking Y component values of corresponding positions from Y pixel values, and calculating RGB component values of other pixels in the group through one pixel;
calculating the maximum value in the R, G and B components of a certain pixel in the pixel group, and recording the index of the maximum value;
calculating histogram statistics of RGB components and maximum components in the pixel group according to the RGB component values and the recorded maximum component index;
and performing traversal accumulation or merging on the histograms of each group of pixel groups, and calculating RGB of the whole image and RGB maximum component histogram statistics.
Preferably, the calculating R, G, B component values of all pixel points in each pixel group specifically includes:
R＝Y+2*(1-Wr)*V
B＝Y+2*(1-Wb)*U
G＝Y-2*(Wr*(1-Wr)*v)/(1-Wb-Wr))
-2*(Wb*(1-Wb)*U)/(1-Wb-Wr))
wherein Wr，Wg，WbAre the weights of the three components R, G, B, respectively, and Wr+Wg+Wb＝1，
According to the color gamut, a conversion standard formula and a corresponding YUV sampling format, taking the value of each pixel point actually by Y, taking the same value in a group by UV, and calculating the R, G and B component values of a group of pixel points;
defining a float type two-dimensional array, storing RGB component values, dividing the array into 3 vectors, wherein the size of each vector is the number of image pixels: float pRGB [3] [ width height ], width and height are defined as the width and height of the image, and each group of pixel values is calculated and stored in the corresponding position.
Preferably, the number of pixels in a group is N, N is 4 when YUV420 samples are input, and N is 2 when YUV422 samples are input; each pixel within a group is represented as: p1＝(R1，G1，B1)…Pn＝(Rn，Gn，Bn)；
The Y component of each pixel within a group can be expressed as: y is1…Yn；
Only one group of UV in a group can be directly denoted as U, V.
Preferably, the maximum value of the components of a certain pixel R, G, B in the pixel group is calculated as: calculating the RGB component value of one pixel in the group, calculating the RGB components of other pixel points through the same difference value to obtain the maximum of R, G and B3 components of the pixel points in a UV group, only calculating the maximum of the first pixel,
the method for calculating and storing the rgbMax component comprises the following steps: setting Idx as the maximum vector index of rgbMax, setting index value 0 to point to the component of image pixel R, index value 1 to point to the component of image pixel G, and index value 2 to point to the component of image pixel B, assuming that the maximum component of one pixel is calculated to be R, then the Idx index value is calculated to be 0, then the indexes of rgbMax of other pixels in the group are also 0, and only one value is stored for one pixel in the group.
And setting pixelIdx as the storage position of the pixel point in the memory, the calculation method is that firstly, which of R and G components is larger is compared, and the larger is compared with B, so that the maximum index component is calculated, namely:
Idx＝pRGB[0][pixelIdx]＞pRGB[1][pixelldx]？0：1；
Idx＝pRGB[Idx][pixelIdx]＞pRGB[2][pixelIdx]？Idx：2；
defining a unit8_ t prgbMaxIdx [ UVwidth UVheight ] structure, and storing maxRGB component indexes corresponding to each group of UV, wherein UVwidth and UVheight are width and height of image UV.
Preferably, histogram statistics of RGB3 components are calculated according to a pRGB [3] [ width × height ] structure, histogram statistics of rgbMax components are calculated according to prgbMaxIdx, the histogram statistics are calculated in units of UV groups, the UV group pixel widths are set to uvW and uvH, 4 pixels (2 × 2) are calculated at a time in a source image format of YUV420, uvW is 2, uvH is 2, and in a YUV422 mode, 2 pixels (2 × 1) are calculated at a time, uvW is 2, and uvH is 1; due to the YUV characteristic, in a group of UV-shared pixel groups, a certain component of a certain pixel is the largest, and then other pixels are also the largest of the same component, and an rgbMax component in a group of pixels is a histogram of the certain component.
The invention has the following beneficial effects:
(1) the dynamic HDR image feature extraction method can simplify the calculation step of extracting the image feature value, reduce the data conversion times and reduce the whole system bandwidth load.
(2) The dynamic HDR image feature extraction method reduces unnecessary calculation, reduces calculation complexity and improves the utilization rate of the calculation resources of the whole equipment.
(3) Based on the characteristics, the extraction and the encoding of the 4K/8K HDR vivid dynamic metadata can be realized on the traditional server + GPU hardware architecture, and a high-density and ultra-high-definition HDR encoding device is realized.
Drawings
FIG. 1 is a schematic diagram of YUV format sampling;
FIG. 2 is a flowchart illustrating steps of a dynamic HDR image feature extraction method according to an embodiment of the present invention;
fig. 3 is a schematic diagram of grouping YUV format image pixels according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 2, a flowchart of steps of a dynamic HDR image feature extraction method according to an embodiment of the present invention is shown, including the following steps:
inputting a YUV image, grouping pixel points of the input image, and recording all pixel points sharing a group of UV as a pixel group;
traversing each pixel group, calculating R, G, B component values of all pixel points in each pixel group, sharing UV values during calculation, taking Y component values of corresponding positions from Y pixel values, and calculating RGB component values of other pixels in the group through one pixel;
calculating the maximum value in the R, G and B components of a certain pixel in the pixel group, and recording the index of the maximum value;
calculating histogram statistics of RGB components and maximum components in the pixel group according to the RGB component values and the recorded maximum component index;
and performing traversal accumulation or merging on the histograms of each group of pixel groups, and calculating RGB of the whole image and RGB maximum component histogram statistics.
In a specific application example, referring to fig. 3, when grouping the input image pixels, a group of UV may correspond to multiple pixels, for example, a group of YUV422 corresponds to 2 pixels, and a group of YUV420 corresponds to 4 pixels. In the encoding distribution, transmission domain, the image uses YUV color space, wherein the YUV420 format is used in most cases, and the YUV422 format is used in a small amount. In this format, multiple pixels share a set of UV. According to the characteristics, the images are grouped according to UV, RGB calculation and statistical data are firstly carried out on pixels in the groups, then RGB values and statistical data of the whole image are calculated, and according to the correlation characteristics of YUV and RGB, the calculation amount and the storage space can be saved.
In a specific application example, the grouped pixel groups are independent from each other, and are suitable for parallel computing processing, such as parallel processing of each pixel group by using CUDA (computer Unified Device Architecture) or openMP (Open Multiprocessing). For the conversion from YUV to RGB, a standard calculation formula can be set for calculation.
The RGB conversion YUV standard formula is as follows:
1)Y＝Wr*R+Wg*G+Wb*B
2)U＝0.5*(B-Y)/(1-Wb)
3)V＝0.5*(R-Y)/(1-Wr)
4)Wr+Wg+Wb＝1
wherein Wr，Wg，WbAre the weights of the three components R, G, B, respectively, and Wr+Wg+Wb1. The weighted values are different under different color gamut standards, but all satisfy the above conditions.
E.g. under BT709 color gamut standard, Wr＝0.2126，Wg＝0.7152，Wb＝0.0722。
BT2020 color gamut standard, Wr＝0.2627，Wg＝0.678，Wb＝0.0593。
The method can be derived according to an RGB conversion YUV formula, wherein R and B are calculated by 2) and 3), and G can be obtained by replacing R and B in 1), namely calculating R, G, B component values of all pixel points in each pixel group specifically comprises the following steps:
R＝Y+2*(1-wr)*V
B＝Y+2*(1-Wb)*U
G＝Y-2*(Wr*(1-Wr)*V)/(1-Wb-Wr))
-2*(Wb*(1-Wb)*U)/(1-Wb-Wr))
further we find BT2020 gamut:
R＝Y+1.4746*V
G＝Y-0.16455*U-0.57135*V
B＝Y+1.8814*U
yield BT709 color gamut:
R＝Y+1.5748*V
G＝Y-0.1873*U-0.4681*V
B＝Y+1.8556*U
calculating the RGB value in the group, wherein Y takes the actual value of each pixel point, and UV takes the same value in the group to calculate the R, G and B component values of a group of pixel points according to the color gamut, a conversion standard formula and a corresponding YUV sampling format;
defining a float type two-dimensional array, storing RGB component values, dividing the array into 3 vectors, wherein the size of each vector is the number of image pixels: float pRGB [3] [ width height ], width and height are defined as the width and height of the image, and each group of pixel values is calculated and stored in the corresponding position.
According to a calculation formula from YUV to RGB, setting the number of pixels in a group as N, wherein N is 4 when YUV420 is input and N is 2 when YUV422 is input; each pixel within a group can be represented as: p1＝(R1，G1，B1)...Pn＝(Rn，Gn，Bn)；
The Y component of each pixel within a group can be expressed as: y is1...Yn；
Only one group of UV in a group can be directly denoted as U, V;
taking BT2020 as an example:
R1＝Y1+1.4746*V
G1＝Y1-0.16455*U-0.57135*V
B1＝Y1+1.8814*U
Rn＝Yn+1.4746*V
Gn＝Yn-0.16455*U-0.57135*V
Bn＝Yn+1.8814*U
the formula is calculated from R, G and B, and since UV is the same, it is only necessary to calculate (R)1，G1，B1)，(Rn，Gn，Bn) Can be prepared from (R)1，G1，B1) And the difference between the Y values.
Rn＝R1+(Yn-Y1)
Gn＝G1+(Yn-Y1)
Bn＝B1+(Yn-Y1)
So only the RGB value of one pixel point in the group is calculated, and then n-1Y pixels are calculatednAnd Y1The RGB values of all pixels in the group can be calculated by the difference value of (3).
Through the specific application example, according to the conversion formula of YUV and RGB, the same UV is shared in a group of pixels, the RGB component value of a certain pixel in the group is calculated, and the RGB component values of other pixels in the group can be obtained through the difference value of Y among the pixels. Thereby saving a portion of the computational effort.
In a specific application example, the maximum value of the components of a certain pixel R, G, B in a pixel group is calculated as: the RGB component value of one pixel in the group is calculated, when the RGB components of other pixels are calculated through the same difference value to obtain the maximum of the R, G and B3 components of the pixels in a UV group, only the maximum of the first pixel is calculated (according to the calculation formula, the same UV in a group of pixels is used, the weight is constant, each pixel in the group has only different Y values, and the RGB component values of other pixels can be calculated through the Y difference value).
The method for calculating and storing the rgbMax component comprises the following steps: setting Idx as the maximum vector index of rgbMax, setting index value 0 to point to the component of image pixel R, index value 1 to point to the component of image pixel G, and index value 2 to point to the component of image pixel B, assuming that the maximum component of one pixel is calculated to be R, then the Idx index value is calculated to be 0, then the indexes of rgbMax of other pixels in the group are also 0, and only one value is stored for one pixel in the group.
And setting pixelIdx as the storage position of the pixel point in the memory, the calculation method is that firstly, which of R and G components is larger is compared, and the larger is compared with B, so that the maximum index component is calculated, namely:
Idx＝pRGB[0][pixelIdx]＞pRGB[1][pixelIdx]？0：1；
Idx＝pRGB[Idx][pixelIdx]＞pRGB[2][pixelIdx]？Idx：2；
defining a unit8_ t prgbMaxIdx [ UVwidth UVheight ] structure, and storing maxRGB component indexes corresponding to each group of UV, wherein UVwidth and UVheight are width and height of image UV.
In a specific application example, histogram statistics of 3 RGB components are calculated according to a pRGB [3] [ width × height ] structure, histogram statistics of 3 RGB components are calculated according to prgbMaxIdx, the histogram statistics are calculated by taking UV grouping as a unit, the widths and heights of UV grouping pixels are set to uvW and uvH, 4 pixels (2x2) are calculated at a time under a source image format of YUV420, uvW is 2, uvH is 2, and 2 pixels (2x1) are calculated at a time under a YUV422 mode, uvW is 2, and uvH is 1; due to the YUV characteristic, in a group of UV-shared pixel groups, a certain component of a certain pixel is the largest, and then other pixels are also the largest of the same component, and an rgbMax component in a group of pixels is a histogram of the certain component.
By the above application example, in a pixel group sharing a group of UV, the maximum components of R, G, and B are consistent, so that the RGB components of only one of the pixels can be calculated, and which component the maximum component value of a group of pixels points to can be recorded by recording the component index. Therefore, the maximum value of the RGB components can be saved from being repeatedly calculated for many times, and the storage space can be saved. The maximum computation amount (the maximum value is computed once by 4 pixels) is saved by 75 percent at most, the maximum rgb component value of the float type is stored, 4 bytes are required, the storage index only needs 1 byte, one pixel is stored in one group, and the minimum storage space occupied by the pixel can be 1/16.
Taking YUV420 as an example, traversing the UV packets, and calculating the coordinate (x, y) of a certain UV packet as follows;
let histoRGB [3] [1025] store the histogram of RGB3 components, and histoMax [1025] store the histogram of the RGB largest component and initialize to 0.
Calculating the statistical value of each component of 4 pixel points at a time:
let rgbIdx0 ═ x UVwidth < 1+ y < 1;
rgbid idx1 ═ rgbid idx0+ 1;
rgbIdx2＝rgbIdx0+width；
rgbIdx3＝rgbIdx1+width；
let rgbMaxIdx ═ x × UVwidth + y; the current rgbMax component maximum index is
prgbMaxIdx[rgbMaxIdx]。
Traversing RGB components, idx is the component index (R ═ 0, G ═ 1, B ═ 2):
the specific flow of calculating each component value of each pixel (taking 10bit quantization as an example) is as follows:
1) the quantized value at a certain component of all pixels in the group is calculated (converted from a floating point number of 0-1 to an integer number of 0-1024).
V0＝pRGB[idx][rgbIdx0]*1024；
V1＝pRGB[idx][rgbIdx1]*1024；
V2＝pRGB[idx][rgbIdx2]*1024；
V3＝pRGB[idx][rgbIdx3]*1024。
2) A histogram of a certain component of RGB within the group is calculated (counting the statistical number of a certain value of each component of 4 pixels within the group, idx being a component).
histoRGB[idX][V0]++；histoRGB[idx][V1]++；
histoRGB[idx][V2]++；histoRGB[idx][V3]++；
3) A histogram of the rgbMax components within the group is calculated. (when 4 pixel point values of R, G and B3 components in the group are traversed, if the recorded regrouped pixel rgbMax index is consistent with the currently traversed component, the 4 pixel values of the component are used for histogram statistics).
If the intra-group RGB maximum index is equal to the current RGB index, then
histoMax[V0]++；histoMax[V1]++；
histoMax[V2]++；histoMax[V3]++。
4) Rgb histogram and maximum component histogram of the whole image are calculated.
And traversing each pixel group, and calculating the histogram statistics of all pixels in the image by adopting a sequential accumulation mode or a pairwise merging mode according to the characteristics of the calculation equipment.
It is to be understood that the exemplary embodiments described herein are illustrative and not restrictive. Although one or more embodiments of the present invention have been described with reference to the accompanying drawings, it will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the spirit and scope of the present invention as defined by the following claims.

Claims (5)

1. A dynamic HDR image feature extraction method is characterized by comprising the following steps:
inputting a YUV image, grouping pixel points of the input image, and recording all pixel points sharing a group of UV as a pixel group;
traversing each pixel group, calculating R, G, B component values of all pixel points in each pixel group, sharing UV values during calculation, taking Y component values of corresponding positions from Y pixel values, and calculating RGB component values of other pixels in the group through one pixel;
calculating the maximum value in the R, G and B components of a certain pixel in the pixel group, and recording the index of the maximum value;
calculating histogram statistics of RGB components and maximum components in the pixel group according to the RGB component values and the recorded maximum component index;
and performing traversal accumulation or merging on the histograms of each group of pixel groups, and calculating RGB of the whole image and RGB maximum component histogram statistics.
2. The method of extracting features of a dynamic HDR image as claimed in claim 1, wherein calculating R, G, B component values of all pixels in each pixel group specifically comprises:
R＝Y+2*(1-Wr)*V
B＝Y+2*(1-Wb)*U
G＝Y-2*(Wr*(1-Wr)*V)/(1-Wb-Wr))-2*(Wb*(1-Wb)*U)/(1-Wb-Wr))
wherein Wr，Wg，WbAre the weights of the three components R, G, B, respectively, and Wr+Wg+Wb＝1，
According to the color gamut, a conversion standard formula and a corresponding YUV sampling format, taking the value of each pixel point actually by Y, taking the same value in a group by UV, and calculating the R, G and B component values of a group of pixel points;
defining a float type two-dimensional array, storing RGB component values, dividing the array into 3 vectors, wherein the size of each vector is the number of image pixels: float pRGB [3] [ width height ], width and height are defined as the width and height of the image, and each group of pixel values is calculated and stored in the corresponding position.
3. The method as claimed in claim 2, wherein the number of pixels in a group is N, N is 4 when YUV420 samples are input, and N is 2 when YUV422 samples are input; each pixel within a group is represented as: p1＝(R1，G1，B1)...Pn＝(Rn，Gn，Bn)；
The Y component of each pixel within a group can be expressed as: y is1...Yn；
Only one group of UV in a group can be directly denoted as U, V.
4. The method of extracting features of a dynamic HDR image as claimed in claim 2, wherein the maximum value of R, G, B components of a certain pixel in the pixel group is calculated as: calculating the RGB component value of one pixel in the group, calculating the RGB components of other pixel points by the same difference value to obtain the maximum of R, G and B3 components of the pixel points in a UV group, only calculating the maximum of the first pixel,
the method for calculating and storing the rgbMax component comprises the following steps: setting Idx as the maximum vector index of rgbMax, setting an index value 0 to point to the R component of an image pixel, setting an index value 1 to point to the G component of the image pixel, and setting an index value 2 to point to the B component of the image pixel, assuming that the maximum component of one pixel is calculated to be R, the Idx index value is calculated to be 0, the indexes of the rgbMax of other pixels in the group are also 0, and only one value is stored in one group of pixels;
and setting pixelIdx as the storage position of the pixel point in the memory, the calculation method is that firstly, which of R and G components is larger is compared, and the larger is compared with B, so that the maximum index component is calculated, namely:
Idx＝pRGB[0][pixelIdx]＞pRGB[1][pixelIdx]？0：1；
Idx＝pRGB[Idx][pixelIdx]＞pRGB[2][pixelIdx]？Idx：2；
defining a unit8_ t prgbMaxIdx [ UVwidth UVheight ] structure, and storing maxRGB component indexes corresponding to each group of UV, wherein UVwidth and UVheight are width and height of image UV.
5. The method as claimed in claim 2, wherein the method calculates histogram statistics of RGB3 components according to pRGB [3] [ width height ] structure, calculates histogram statistics of rgbMax components according to prgbMaxIdx, calculates histogram statistics in units of UV groups, sets UV group pixel width as uvW and uvH, calculates 4 pixels (2x2) at a time in YUV420 format, uvW is 2, uvH is 2, calculates 2 pixels (2x1) at a time in YUV422 mode, uvW is 2, uvH is 1; due to the YUV characteristic, in a group of UV-shared pixel groups, a certain component of a certain pixel is the largest, and then other pixels are also the largest of the same component, and an rgbMax component in a group of pixels is a histogram of the certain component.
CN202110781871.1A 2021-07-09 2021-07-09 Dynamic HDR image feature extraction method Active CN113591878B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110781871.1A CN113591878B (en) 2021-07-09 2021-07-09 Dynamic HDR image feature extraction method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110781871.1A CN113591878B (en) 2021-07-09 2021-07-09 Dynamic HDR image feature extraction method

Publications (2)

Publication Number Publication Date
CN113591878A true CN113591878A (en) 2021-11-02
CN113591878B CN113591878B (en) 2024-07-23

Family

ID=78246910

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110781871.1A Active CN113591878B (en) 2021-07-09 2021-07-09 Dynamic HDR image feature extraction method

Country Status (1)

Country Link
CN (1) CN113591878B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116883323A (en) * 2023-06-16 2023-10-13 中国医学科学院肿瘤医院 Multiple immunofluorescence image analysis system based on computer vision

Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20030234795A1 (en) * 2002-06-24 2003-12-25 Samsung Electronics Co., Ltd. Apparatus and method for converting of pixels from YUV format to RGB format using color look-up tables
US20050213125A1 (en) * 2002-08-19 2005-09-29 Paul Reed Smith Guitars, Limited Partnership Method of color accentuation with compensation and adjustment
US20050249283A1 (en) * 2004-05-07 2005-11-10 Canon Kabushiki Kaisha Image coding apparatus and image decoding apparatus and their control methods, and computer program and computer-readable storage medium
CN101340511A (en) * 2008-08-07 2009-01-07 中兴通讯股份有限公司 Adaptive video image enhancing method based on lightness detection
US20100060795A1 (en) * 2008-09-08 2010-03-11 Samsung Electronics Co., Ltd. Color interpolation device and color interpolation method
EP2528319A1 (en) * 2011-05-23 2012-11-28 Alcatel Lucent Image data compressing and decompressing methods and devices
CN108460729A (en) * 2018-01-17 2018-08-28 东莞市爱协生智能科技有限公司 A kind of computer readable storage medium and the Image Reversal device using the medium
CN108683899A (en) * 2018-05-16 2018-10-19 东南大学 A kind of color space conversion optimization method of Embedded image processing system
CN108781290A (en) * 2016-03-07 2018-11-09 皇家飞利浦有限公司 HDR videos are coded and decoded
CN108781246A (en) * 2016-03-14 2018-11-09 皇家飞利浦有限公司 Saturation degree processing for dynamic range mapping is specified
CN111105356A (en) * 2019-12-26 2020-05-05 Tcl华星光电技术有限公司 Image processing method, image processing apparatus, and computer-readable storage medium

Patent Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20030234795A1 (en) * 2002-06-24 2003-12-25 Samsung Electronics Co., Ltd. Apparatus and method for converting of pixels from YUV format to RGB format using color look-up tables
US20050213125A1 (en) * 2002-08-19 2005-09-29 Paul Reed Smith Guitars, Limited Partnership Method of color accentuation with compensation and adjustment
US20050249283A1 (en) * 2004-05-07 2005-11-10 Canon Kabushiki Kaisha Image coding apparatus and image decoding apparatus and their control methods, and computer program and computer-readable storage medium
CN101340511A (en) * 2008-08-07 2009-01-07 中兴通讯股份有限公司 Adaptive video image enhancing method based on lightness detection
US20100060795A1 (en) * 2008-09-08 2010-03-11 Samsung Electronics Co., Ltd. Color interpolation device and color interpolation method
EP2528319A1 (en) * 2011-05-23 2012-11-28 Alcatel Lucent Image data compressing and decompressing methods and devices
CN108781290A (en) * 2016-03-07 2018-11-09 皇家飞利浦有限公司 HDR videos are coded and decoded
CN108781246A (en) * 2016-03-14 2018-11-09 皇家飞利浦有限公司 Saturation degree processing for dynamic range mapping is specified
CN108460729A (en) * 2018-01-17 2018-08-28 东莞市爱协生智能科技有限公司 A kind of computer readable storage medium and the Image Reversal device using the medium
CN108683899A (en) * 2018-05-16 2018-10-19 东南大学 A kind of color space conversion optimization method of Embedded image processing system
CN111105356A (en) * 2019-12-26 2020-05-05 Tcl华星光电技术有限公司 Image processing method, image processing apparatus, and computer-readable storage medium

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
YANJIE LIU，等: "A Fast Fusion Method for Multi-exposure Image in YUV Color Space", 《IEEE》, pages 18346555 *
崔唱: "新型图像压缩及优化算法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》, no. 12, pages 136 - 130 *
张芸，等: "几种基于颜色模型的彩色图像黑白化算法", 南昌工程学院学报, vol. 36, no. 04, pages 35 - 38 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116883323A (en) * 2023-06-16 2023-10-13 中国医学科学院肿瘤医院 Multiple immunofluorescence image analysis system based on computer vision

Also Published As

Publication number Publication date
CN113591878B (en) 2024-07-23

Similar Documents

Publication Publication Date Title
US11212539B2 (en) Efficient lossless compression of captured raw image information systems and methods
US11445201B2 (en) Video encoding and decoding method, device, and system, and storage medium
KR102358368B1 (en) Method and device for encoding high dynamic range pictures, corresponding decoding method and decoding device
WO2020119449A1 (en) Chroma block prediction method and device
WO2020135357A1 (en) Data compression method and apparatus, and data encoding/decoding method and apparatus
KR20180021869A (en) Method and device for encoding and decoding HDR color pictures
KR102617258B1 (en) Image processing method and apparatus
US10509982B2 (en) On-camera image processing based on image luminance data
CN107846589B (en) Image compression method based on local dynamic quantization
KR20220043158A (en) Image processing method, apparatus, electronic device, and storage medium
CN111491149B (en) Real-time image matting method, device, equipment and storage medium based on high-definition video
WO2021073304A1 (en) Image processing method and apparatus
CN113068034B (en) Video encoding method and device, encoder, equipment and storage medium
CN107087163A (en) A kind of coding method of lifting HDR Subjective video qualities
CN113591878B (en) Dynamic HDR image feature extraction method
CN114223209A (en) Adaptive image data linearization for HDR image sensors
CN114998122A (en) Low-illumination image enhancement method
WO2022141515A1 (en) Video encoding method and device and video decoding method and device
US20230396810A1 (en) Hierarchical audio/video or picture compression method and apparatus
CN113542864B (en) Video splash screen area detection method, device and equipment and readable storage medium
CN111406404B (en) Compression method, decompression method, system and storage medium for obtaining video file
CN116156140A (en) Video display processing method, device, computer equipment and storage medium
CN116320446A (en) Video encoding method and device, electronic equipment and computer medium
CN116320465A (en) Video compression and transmission method, device, gateway and storage medium
AU2020101803A4 (en) Method for Optimizing the Coding Block-Level Lagrange Multiplier of Longitude-Latitude Image

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant