WO2022267136A1 - 超分辨率图像重构方法及装置 - Google Patents

超分辨率图像重构方法及装置 Download PDF

Info

Publication number
WO2022267136A1
WO2022267136A1 PCT/CN2021/107498 CN2021107498W WO2022267136A1 WO 2022267136 A1 WO2022267136 A1 WO 2022267136A1 CN 2021107498 W CN2021107498 W CN 2021107498W WO 2022267136 A1 WO2022267136 A1 WO 2022267136A1
Authority
WO
WIPO (PCT)
Prior art keywords
pixels
pixel
row
column
adjacent
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.)
Ceased
Application number
PCT/CN2021/107498
Other languages
English (en)
French (fr)
Inventor
刘金风
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.)
TCL China Star Optoelectronics Technology Co Ltd
Original Assignee
TCL China Star Optoelectronics 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 TCL China Star Optoelectronics Technology Co Ltd filed Critical TCL China Star Optoelectronics Technology Co Ltd
Priority to US17/429,950 priority Critical patent/US12462338B2/en
Publication of WO2022267136A1 publication Critical patent/WO2022267136A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00—Geometric image transformations in the plane of the image
    • G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
    • G06T3/4053—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00—Geometric image transformations in the plane of the image
    • G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
    • G06T3/4023—Scaling of whole images or parts thereof, e.g. expanding or contracting based on decimating pixels or lines of pixels; based on inserting pixels or lines of pixels

Definitions

  • the present application relates to the field of display technology, and in particular to a super-resolution image reconstruction method and device.
  • 8K display panels With the continuous advancement of display technology, ultra-large-sized, ultra-high-resolution display panels have been released. For example, after 4K display panels, 8K display panels are gradually mass-produced. However, limited by network bandwidth, camera equipment, and video codec technology, 8K display panels have a lot of shortages in film sources. Previous 8K TVs usually directly stretch and match video signals smaller than 8K resolution to 8K resolution, for example, each pixel in a 4K video signal is stretched 2*2 times to correspond to 4 pixels, but the display effect of such 8K resolution is extremely poor.
  • each pixel in the 4K video signal can only correspond to one pixel in the 8K resolution display panel, for example, the image is regarded as a matrix, and the value of the row and column of the matrix corresponds to a pixel, and the matrix elements are For the gray value of the pixel, the pixel in the i-th row and j-column of the original image matrix of the 4K video is assigned to [2i-1, 2j-1] of the 8K resolution target matrix (both i and j are Positive integer), to achieve the effect of doubling the original image matrix to the target matrix, and then use the adjacent replacement algorithm, bilinear interpolation algorithm, cubic interpolation and other algorithms to fill the blank pixels in the 8K resolution target matrix.
  • these algorithms either have problems such as jagged, blurred, and grainy images due to insufficient consideration of the internal relationship between adjacent pixels, or the calculation process is too complex and inefficient, resulting in poor real-time performance.
  • the current super-resolution image reconstruction methods either have problems such as jagged, blurred, and grainy images due to insufficient consideration of the internal relationship between adjacent pixels, or the calculation process is too complex and inefficient, resulting in poor real-time performance. .
  • the present application provides a super-resolution image reconstruction method and device.
  • the present application provides a super-resolution image reconstruction method, including:
  • the original image includes a plurality of pixels distributed in an array; wherein, each row of pixels includes a plurality of row pixel groups composed of continuous pixels, and each column of pixels includes a plurality of continuous pixels The column pixel group;
  • step S2 according to the visual relationship between the pixels in each row of pixel groups, the gray value corresponding to the newly added row of pixels between two adjacent pixels in each row of pixels is acquired Specifically include:
  • step S2 according to the visual relationship between the pixels in each column of pixel groups, the gray value corresponding to the newly added column of pixels between two adjacent pixels in each column of pixels is obtained , including:
  • the preset visible condition includes: the gray values of the pixels located between the row position of each pixel and the row position of the visible row pixels corresponding to each pixel are smaller than each pixel's grayscale value and the grayscale value of each pixel in the visible row.
  • the preset visual condition includes: the gray value of the pixels between the column position of each pixel and the column position of the visible column corresponding to each pixel is smaller than that of each pixel and the gray value of each pixel in the visible column.
  • step S22 specifically includes:
  • each pixel and each of the corresponding visible row pixels select the visual functional relationship associated with two adjacent pixels in each row of pixels;
  • the visual functional relationship associated with two pixels is a visual functional relationship that intersects with the row positions between two adjacent pixels;
  • step S22' specifically includes:
  • each pixel and each of the corresponding visible column pixels select the visual functional relationship associated with two adjacent pixels in each column of pixels;
  • the visual functional relationship associated with two adjacent pixels is a visual functional relationship that intersects with the column positions between two adjacent pixels;
  • step S223 specifically includes:
  • any row position between two adjacent pixels and all the visual functional relationships associated with the two adjacent pixels, obtain a plurality of initial gray values corresponding to the newly added row pixels at any one row position degree value;
  • the average value of the multiple initial gray values is used as the gray value corresponding to the newly added row pixel at any one row position.
  • step S223' specifically includes:
  • any column position between two adjacent pixels and all the visual functional relationships associated with the two adjacent pixels, obtain a plurality of initial gray values corresponding to the newly added column pixels at any one column position degree value;
  • the average value of the multiple initial grayscale values is used as the grayscale value corresponding to the newly added column pixel at any one column position.
  • step S3 specifically includes: combining the gray value of the pixel in the new row between the adjacent two pixels in the two adjacent rows with the gray value of the pixel in the adjacent two columns between the two adjacent pixels The average value of the grayscale values of the newly added column pixels is used as the grayscale value corresponding to the newly added central pixel.
  • the number of the newly added row of pixels inserted between two adjacent pixels of each row of pixels is the same, and/or, the number of the newly added row of pixels inserted between adjacent two pixels of each column of pixels same amount.
  • the present application provides a super-resolution image reconstruction device, including:
  • the acquisition module is used to acquire the original image to be reconstructed, and the original image includes a plurality of pixels distributed in an array; wherein, each row of pixels includes a plurality of row pixel groups composed of continuous pixels, and each column of pixels includes a plurality of row pixel groups composed of A column pixel group composed of consecutive pixels;
  • a new row and column pixel module is used to obtain the gray value corresponding to the newly added row pixel between two adjacent pixels in each row of pixels according to the visual relationship between the pixels in each row pixel group, And, according to the visual relationship between the pixels in each column of pixel groups, determine the gray value corresponding to the newly added column of pixels between two adjacent pixels in each column of pixels;
  • Adding a central pixel module configured to determine the pixels in the new row between the pixels in two adjacent rows and the pixels in two adjacent columns according to the pixels in the newly added row and the pixels in the newly added columns The gray value corresponding to the newly added center pixel in the surrounding area of the newly added column of pixels between two adjacent pixels; wherein, the adjacent two pixels of the two adjacent rows are the same as the The adjacent two pixels of overlap;
  • an inserting module configured to insert the newly added row of pixels between two adjacent pixels in each row of pixels, insert the newly added column of pixels between two adjacent pixels in each column of pixels, and The central pixel is inserted into the surrounding area of the newly added row of pixels and the column of pixels, so as to obtain a reconstructed super-resolution image corresponding to the original image.
  • the newly added row and column pixel module specifically includes:
  • a row pixel visual relationship unit configured to determine at least one visible row pixel of each pixel in the row pixel group according to the gray value of each pixel and the row position of each pixel in the row pixel group; wherein, Each pixel satisfies a preset visibility condition between at least one visible line pixel corresponding to each pixel;
  • a new row pixel gray value acquisition unit is used to obtain the gray value and row position of each pixel in each row of pixels according to the gray value and row position of the visible row pixels corresponding to each pixel.
  • the gray value corresponding to the newly added row pixel at any row position between two adjacent pixels.
  • the newly added row and column pixel module specifically includes:
  • a column pixel visual relationship unit configured to determine at least one visible column pixel of each pixel in the column pixel group according to the gray value of each pixel and the column position of each pixel in the pixel column; wherein, Each pixel satisfies a preset visual condition with at least one pixel in the visible column corresponding to each pixel;
  • a new column pixel gray value acquisition unit is used to obtain the gray value and column position of each pixel in each column of pixels according to the gray value and column position of each pixel corresponding to the visible column pixel.
  • the gray value corresponding to the new column pixel at any column position between two adjacent pixels.
  • the newly added row pixel gray value acquisition unit is specifically used for:
  • each pixel According to the grayscale value and row position of each pixel, and the grayscale value and row position of the visible row pixels corresponding to each pixel, the distance between each pixel and each corresponding visible row pixel is respectively established.
  • each pixel and each corresponding visible row pixel select the visual functional relationship associated with two adjacent pixels in each row of pixels;
  • the visual functional relationship associated with the pixel is a visual functional relationship that intersects with the row positions between two adjacent pixels;
  • the gray value corresponding to the new row pixel at any one row position between the two adjacent pixels is acquired.
  • the new column pixel gray value acquisition unit is specifically used for:
  • each pixel According to the grayscale value and column position of each pixel, and the grayscale value and column position of the visible column pixels corresponding to each pixel, the distance between each pixel and each corresponding visible column pixel is respectively established.
  • each pixel and each corresponding visible column pixel select the visual functional relationship associated with two adjacent pixels in each column of pixels;
  • the visual functional relationship associated with the pixels is a visual functional relationship that intersects with the column positions between two adjacent pixels;
  • the gray value corresponding to the newly added column pixel at the position of any column between the two adjacent pixels is acquired.
  • the newly-added row pixel gray value acquisition unit is specifically configured to obtain the new value of any row position between two adjacent pixels according to the visual function relationship associated with two adjacent pixels.
  • the gray value corresponding to the row-by-row pixel further includes:
  • any row position between two adjacent pixels and all the visual functional relationships associated with the two adjacent pixels, obtain a plurality of initial gray values corresponding to the newly added row pixels at any one row position degree value;
  • the average value of the multiple initial gray values is used as the gray value corresponding to the newly added row pixel at any one row position.
  • the newly added column pixel gray value acquisition unit is specifically configured to acquire the new value of any one column position between two adjacent pixels according to the visual function relationship associated with two adjacent pixels. Add the gray value corresponding to the pixel, further including:
  • any column position between two adjacent pixels and all the visual functional relationships associated with the two adjacent pixels, obtain a plurality of initial gray values corresponding to the newly added column pixels at any one column position degree value;
  • the average value of the multiple initial grayscale values is used as the grayscale value corresponding to the newly added column pixel at any one column position.
  • the newly added central pixel module is specifically used for:
  • the gray value of the newly added row pixel between two adjacent pixels of the two adjacent rows, and the gray value of the newly added column pixel between two adjacent pixels of two adjacent columns The average value of the values is used as the gray value corresponding to the newly added center pixel.
  • the present application provides an electronic device, including a memory, a processor, and a computer program stored on the memory and operable on the processor.
  • the processor executes the program, the above-mentioned super-resolution is realized.
  • Image reconstruction methods including a processor, a processor, and a computer program stored on the memory and operable on the processor.
  • the present application provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the super-resolution image reconstruction method as described above are implemented.
  • the present application provides a super-resolution image reconstruction method and device.
  • the gray value corresponding to the new pixel that needs to be inserted between every two adjacent pixels in each row or column, and insert the new pixel of the gray value into two adjacent pixels in each row or column,
  • a super-resolution image reconstructed from the original image is obtained.
  • the super-resolution image reconstruction method fully considers the internal relationship between the local pixels of the original image, it can better maintain the detailed information of the image, and the problems of jaggedness, blurring, and graininess are obviously weakened, and the original image is improved.
  • the method of the present invention has simple principle, small amount of calculation, high real-time processing capability, and ensures high-speed signal transmission and high-efficiency operation of the system.
  • FIG. 1 is a schematic flow chart of a super-resolution image reconstruction method provided in an embodiment of the present application
  • FIG. 2 is a histogram of the row position and gray value of any row of pixels in the original image provided by the embodiment of the present application;
  • 3 is a histogram of row positions and gray values of pixels in a row pixel group of the original image provided by the embodiment of the present application;
  • FIG. 5 is a schematic diagram of the process of reconstructing an original image into a super-resolution image provided by an embodiment of the present application
  • FIG. 6 is a schematic structural diagram of a super-resolution image reconstruction device provided by an embodiment of the present application.
  • FIG. 7 is a schematic structural diagram of an electronic device provided by an embodiment of the present application.
  • Fig. 1 is a schematic flow chart of the super-resolution image reconstruction method provided by the embodiment of the present application. As shown in Fig. 1, the embodiment of the present application provides a super-resolution image reconstruction method, including:
  • the original image includes a plurality of pixels distributed in an array; wherein, each row of pixels includes a plurality of row pixel groups composed of continuous pixels, and each column of pixels includes a plurality of columns composed of continuous pixels. group of pixels.
  • continuous means that the pixels in each pixel group are pixel points arranged in sequence in each row or column, that is, each pixel forms a row pixel group with adjacent row pixels, and each pixel forms column pixels with adjacent column pixels. Group.
  • each row of pixels is divided into a plurality of row pixel groups, and each row pixel group is composed of a first preset number of continuous pixels.
  • each row of pixels is divided into It is a plurality of column pixel groups, and each column pixel group is composed of a second preset number of continuous pixels, so as to facilitate determining the relationship between each pixel and its adjacent pixels in each row pixel group and each column pixel group in subsequent steps
  • new row pixels or new column pixels with the best gray value are filled between two adjacent pixels.
  • the number of consecutive pixels included between each row pixel group, between each column pixel group, or between a row pixel group and a column pixel group may be the same or different, so the first preset
  • the set number and the second preset number are not constant values, and the first preset number and the second preset number are not necessarily the same, and the first preset can be set according to the key reconstruction part and the non-key reconstruction part of the original image
  • the specific values of the number and the second preset number for example, if the pixel group of the key reconstruction part includes more pixels, the first preset number and the second preset number will be larger, such as 6-10, instead of the key point
  • the pixel groups in the reconstructed part include fewer pixels, and the first preset number and the second preset number are smaller, such as 1-5.
  • each pixel does not necessarily belong to only one row pixel group or one column pixel group, that is, two adjacent pixel groups may include overlapping pixels, as long as the specific pixel points included in each pixel group
  • the gray value of the pixel that has a visual relationship with each pixel is selected in each pixel group as the determination of each adjacent pixel.
  • the black dots are the pixels of the original image
  • the white dots are the new row pixels
  • the rectangles are the new column pixels
  • the triangles are the new center pixels.
  • Figure 5 shows that the original image is doubled Schematic diagram of obtaining a super-resolution image. After inserting a new row of pixels between two adjacent pixels of each row of pixels, and inserting a new row of pixels between two adjacent columns of pixels in each column of pixels, for two adjacent pixels of two adjacent rows and For the four pixels in which two adjacent pixels in two adjacent columns overlap (such as the square dotted frame in Figure 5), there are blank pixels in the surrounding area between the newly added row pixels and the newly added column pixels, so A new central pixel needs to be inserted in this surrounding area.
  • the gray value of the newly added central pixel in this embodiment is determined by the gray value of the two newly added row pixels and newly added column pixels surrounding it.
  • a new row of pixels of the gray value obtained above is inserted between two adjacent pixels of each row of pixels, and a new row of pixels of the gray value obtained above is inserted between two adjacent pixels of each column of pixels.
  • each pixel in the original image needs to The gray value corresponding to the new row pixel and the new column pixel that needs to be inserted between every two adjacent pixels of each row or column, and then determine the overlapping adjacent two row pixels according to the new row pixel and the new column pixel
  • the gray value corresponding to the newly added central pixel in the surrounding area of two adjacent pixels and adjacent two adjacent pixels in two adjacent columns and insert the newly added pixels that meet the corresponding gray value into the adjacent area of each row or column
  • the two pixels, and the newly added central pixel corresponding to the corresponding gray value is inserted into the surrounding area of the newly added row pixel and the newly added column pixel, so as to obtain a super-resolution image reconstructed from the original image.
  • the super-resolution image reconstruction method fully considers the internal relationship between the local pixels of the original image, it can better maintain the detailed information of the image, and the problems of jaggedness, blurring, and graininess are obviously weakened, and the original image is improved.
  • the method of the present invention has simple principle, small amount of calculation, high real-time processing capability, and ensures high-speed signal transmission and high-efficiency operation of the system.
  • step S2 according to the visual relationship between the pixels in each row of pixels, obtaining the gray value corresponding to the newly added row of pixels between two adjacent pixels in each row of pixels specifically includes:
  • S21 Determine at least one visible row pixel of each pixel in the corresponding row pixel group according to the gray value of each pixel and the row position of each pixel in the corresponding pixel row.
  • a preset visible condition is satisfied, and the preset visible condition includes: the row position of each pixel corresponds to the visible row pixel corresponding to each pixel
  • the gray value of the pixels between the row positions is smaller than the gray value of each pixel and the gray value of each visible row pixel.
  • the row position of the pixel is used as the abscissa (the value of the row position can be set by itself, and the row positions 1, 2, 3... 2, 3rd...pixel), the gray value of the pixel is used as the vertical coordinate, and the row position-gray value histogram is drawn.
  • the criterion for judging the visual relationship in the embodiment of the present application is: if two pixels located in the same row are visible, then any row position between the two pixels satisfies the following formula:
  • (Xi, Pi) and (Xj, Pj) are the row position value and gray value of two pixels located in the same row respectively, and (Xm, Pm) is the pixel at any row position between these two pixels The row position and gray value of .
  • the gray value of the pixel in any row position between two pixels with a visual relationship in the same row is smaller than the gray value of these two pixels, that is, the gray value of the pixels in Figure 2 has If the connection between the gray values of the pixels with visual relationship is not blocked by the gray value bars of other pixels, this is taken as the preset visual condition. On the contrary, if the connection between the gray values of two pixels If the line is blocked by the gray value histogram of other pixels between these two pixels, there is no visual relationship between these two pixels.
  • the preset visible condition determine at least one visible row pixel of each pixel in the row pixel group, for example: if the first preset number is 4, Fig. 3 and Fig. 4 are respectively two by 4
  • the row position-gray value histogram of the row pixel group consisting of four consecutive pixels (such as the rectangular dotted line box in Figure 5), the row positions of these four consecutive pixels are respectively row position 1, row position 2, and row position 3 and row position 4.
  • the pixel at row position 1 has a visual relationship with the pixels at row position 2, row position 3, and row position 4 respectively, so the visible pixel of the pixel at row position 1 in the row pixel group is the row position 2.
  • the visible pixels in this row of pixel groups are the pixels at row position 1, row position 2 and row position 4; the visible pixels of row position 4 in this row of pixel groups are row position 1, row position 2 and row Pixel at position 3.
  • the visible pixels of the pixel at row position 1 in the row pixel group are the pixels at row position 2 and row position 3; similarly, the visible pixels of the pixel at row position 2 in the row pixel group
  • the pixels at row position 1 and row position 3; the visible pixels of the pixel at row position 3 in the row pixel group are the pixels at row position 1, row position 2 and row position 4; the pixels at row position 4 are at the row of pixels
  • the visible pixel in the group is the pixel in row position 3.
  • step S22 specifically includes:
  • each pixel According to the grayscale value and row position of each pixel, and the grayscale value and row position of each visible row pixel corresponding to each pixel, respectively establish a possible distance between each pixel and each corresponding visible row pixel Look at the functional relationship.
  • the gray values of the pixels at row position 1, row position 2, row position 3 and row position 4 in Fig. 3 are 240, 160, 140 and 210 respectively, according to the pixel at row position 1 and the pixel at row position 2
  • each pixel and each corresponding visible row pixel select the visual functional relationship associated with two adjacent pixels in each row of pixels;
  • the visual functional relationship associated with the pixels is a visual functional relationship that intersects with the row positions between two adjacent pixels.
  • the visual function relationship of intersection with the row position between two adjacent pixels means that in the histogram, the connection between these two adjacent pixels and the corresponding visible pixel needs to go through this Between two adjacent pixels, the visual functional relationship corresponding to these lines will be used as the visual functional relationship associated with the two adjacent pixels.
  • the visual function relationship P12 between the pixel at row position 1 and the pixel at row position 2 should be selected in Figure 3, row The visual functional relationship P13 between the pixel at position 1 and the pixel at row position 3, and the visual functional relationship P14 between the pixel at row position 1 and the pixel at row position 4, as the relationship between the pixel at row position 1 and the row position 2, and in Figure 4, the visual function relationship P12 between the pixel at row position 1 and the pixel at row position 2 should be selected, and the pixel at row position 1 and the pixel at row position 3
  • the visual functional relationship P13 between is as the visual functional relationship associated with the pixel at row position 1 and the pixel at row position 2.
  • the visual functional relationship P23, and the visual functional relationship P31 between the pixel at row position 3 and the pixel at row position 1 serve as the visual functional relationship associated with row position 2 and row position 3.
  • step S223 specifically includes: according to any row position between two adjacent pixels, and all visual functional relationships associated with the two adjacent pixels, obtain the new row pixel corresponding to any row position multiple initial grayscale values, and the average value of the multiple initial grayscale values is used as the grayscale value corresponding to the newly added row pixel at any row position.
  • step S3 specifically includes:
  • the number of new rows of pixels inserted between two adjacent pixels of each row of pixels is the same, and each column of pixels
  • the number of new columns of pixels inserted between two adjacent pixels is the same, for example, when expanding an original 4K resolution image to an 8K super-resolution image, insert between every adjacent two pixels of each row A new row of pixels, and a new column of pixels is inserted between every two adjacent pixels of each column, instead of not inserting pixels between some adjacent two pixels, but between some adjacent two pixels Insert more than one new pixel between each adjacent pixel, that is, insert the same number of pixels evenly between every two adjacent pixels, so that the display effect of the final super-resolution image is uniform, and there will be no partial area blurring, jaggies and Graininess is strong, and inhomogeneity is very clear in some areas.
  • the distance between the row position of each newly added pixel and the row positions of two adjacent pixels can also be kept consistent, for example, in the same Insert the new pixel adjacent to the midpoint of two pixels.
  • the embodiment of the present application also provides a super-resolution image reconstruction device, including:
  • the acquisition module 601 is used to acquire the original image to be reconstructed, the original image includes a plurality of pixels distributed in an array; wherein, each row of pixels includes a plurality of row pixel groups composed of continuous pixels, and each column of pixels includes a plurality of A column pixel group consisting of consecutive pixels;
  • Adding a row and column pixel module 602 configured to obtain the gray value corresponding to the newly added row pixel between two adjacent pixels in each row of pixels according to the visual relationship between the pixels in each row pixel group , and, according to the visual relationship between the pixels in each row of pixel groups, determine the gray value corresponding to the newly added row of pixels between two adjacent pixels in each row of pixels;
  • Adding a central pixel module 603, configured to determine, according to the newly added row pixels and the newly added column pixels, the newly added row pixels and the two adjacent columns between two adjacent pixels in two adjacent rows The gray value corresponding to the newly added center pixel in the surrounding area of the newly added row of pixels between two adjacent pixels; wherein, the adjacent two pixels of the two adjacent rows are the same as the adjacent two Two adjacent pixels of the column coincide;
  • an inserting module 604 configured to insert the newly added row of pixels between two adjacent pixels in each row of pixels, insert the newly added row of pixels between two adjacent pixels in each column of pixels, and A central pixel is inserted into the surrounding area of the newly added row of pixels and the column of pixels, so as to obtain a super-resolution image corresponding to the reconstruction of the original image.
  • the super-resolution image reconstruction device knows that each pixel in the original image needs to be reconstructed according to the visual relationship between each pixel in the original image and other pixels in the row pixel group or column pixel group.
  • the super-resolution image reconstruction method fully considers the internal relationship between the local pixels of the original image, it can better maintain the detailed information of the image, and the problems of jaggedness, blurring, and graininess are obviously weakened, and the original image is improved.
  • the method of the present invention has simple principle, small amount of calculation, high real-time processing capability, and ensures high-speed signal transmission and high-efficiency operation of the system.
  • Fig. 7 is a schematic structural diagram of an electronic device provided by an embodiment of the present invention.
  • an embodiment of the present invention provides an electronic device, which may include: a processor (processor) 701, a communication interface (Communications Interface) 702 , a memory (memory) 703 and a communication bus 704 , wherein the processor 701 , the communication interface 702 , and the memory 703 communicate with each other through the communication bus 704 .
  • the communication interface 702 can be used for information transmission between the server and the smart TV.
  • the processor 701 can call the logic instructions in the memory 703 to perform the following method, for example, including: S1, acquiring an original image to be reconstructed, the original image includes a plurality of pixels distributed in an array; wherein, each row of pixels includes a plurality of A row pixel group composed of continuous pixels, each column pixel includes a plurality of column pixel groups composed of continuous pixels; S2, according to the visual relationship between the pixels in each row pixel group, obtain two adjacent pixels in each row pixel group The gray value corresponding to the newly-added row pixels between pixels, and, according to the visual relationship between the pixels in each column pixel group, determine the newly-added column between two adjacent pixels in each column of pixels The grayscale value corresponding to the pixel; S3, according to the newly added row pixel and the newly added column pixel, determine the newly added row pixel between the adjacent two adjacent pixels of two adjacent rows and the adjacent two adjacent pixels of two adjacent columns The gray value corresponding to the newly added central pixel in the center of the newly added column pixels between; wherein, the
  • This embodiment also provides a non-transitory computer-readable storage medium, the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions cause the computer to execute the methods provided in the above method embodiments, for example Including: S1. Acquiring the original image to be reconstructed.
  • the original image includes a plurality of pixels distributed in an array; wherein, each row of pixels includes a plurality of row pixel groups composed of continuous pixels, and each column of pixels includes a plurality of row pixel groups composed of continuous pixels.
  • the pixels in each row of pixel groups obtain the gray value corresponding to the newly added row pixel between two adjacent pixels in each row of pixels, and, according to The visual relationship between the pixels in each column pixel group determines the gray value corresponding to the newly added column pixel between two adjacent pixels in each column pixel; S3, according to the newly added row pixel and the newly added column pixel, determine the gray level corresponding to the newly added central pixel in the center of the newly added row pixel between the adjacent two adjacent pixels of the adjacent two rows and the newly added column pixel between the adjacent two adjacent columns of pixels Value; Wherein, the adjacent two pixels of adjacent two rows coincide with the adjacent two pixels of adjacent two columns; S4, insert new row of pixels between the adjacent two pixels in each row of pixels, in each row of pixels Inserting new column pixels between two adjacent pixels in the column pixels, and inserting a center pixel between the newly added row pixels and column pixels, so as to obtain the reconstructed super-resolution image corresponding to the original image.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Image Processing (AREA)

Abstract

本申请提供一种超分辨率图像重构方法及装置，由于该超分辨率图像重构方法充分考虑了原始图像的局部像素之间的内在联系，因此可以较好地保持图像的细节信息，锯齿、模糊、颗粒感等问题明显减弱，改善了原始图像经拉伸后的显示效果。

Description

超分辨率图像重构方法及装置 技术领域
本申请涉及显示技术领域，尤其涉及一种超分辨率图像重构方法及装置。
背景技术
随着显示技术的不断进步，超大尺寸、超高分辨率的显示面板已经面世，例如在4K显示面板之后，8K显示面板也逐渐量产。然而，受网络带宽、摄像设备以及视频编解码技术的限制，8K显示面板在片源方面存在很大不足。之前的8K电视通常是将小于8K分辨率的视频信号直接拉伸匹配到8K分辨率，例如将4K视频信号中的每个像素点拉伸2*2倍来对应8K分辨率显示面板中的4个像素点，但是这样8K分辨率的显示效果极差。
目前虽然还可以将4K视频信号中的每个像素点仅对应8K分辨率显示面板中的一个像素点，例如将图像视为一个矩阵，矩阵的行与列的值对应一个像素点，矩阵元素为该像素点的灰度值，则将4K视频的原始图像矩阵的第i行第j列的像素点赋值到8K分辨率的目标矩阵的【2i-1,2j-1】(i和j均为正整数)，实现将原始图像矩阵扩大一倍到目标矩阵的效果，然后采用邻近取代算法、双线性插值算法、三次内插法等算法填充8K分辨率的目标矩阵中的空白像素点。但是，这些算法要么由于未充分考虑邻近像素之间的内在联系而使得显示画面存在锯齿、模糊、颗粒感明显等问题，要么由于计算过程过于复杂而效率较低，使得实时性差。
技术问题
目前的超分辨率图像重构方法要么由于未充分考虑邻近像素之间的内在联系而使得显示画面存在锯齿、模糊、颗粒感明显等问题，要么由于计算过程过于复杂而效率较低，使得实时性差。
因此，有必要提出一种新的超分辨率图像重构方法，能实现显示 画面清晰，以及计算简单和实时性较好的效果。
技术解决方案
为了解决上述问题，本申请提供一种超分辨率图像重构方法及装置。
第一方面，本申请提供一种超分辨率图像重构方法，包括：
S1、采集待重构的原始图像，所述原始图像包括多个呈阵列分布的像素；其中，每行像素包括多个由连续像素组成的行像素组，每列像素包括多个由连续像素组成的列像素组；
S2、根据每个所述行像素组中的像素之间的可视关系，获取每行像素中的相邻两个像素之间的新增行像素对应的灰度值，以及，根据每个所述列像素组中的像素之间的可视关系，确定每列像素中的相邻两个像素之间的新增列像素对应的灰度值；
S3、根据所述新增行像素和所述新增列像素，确定在相邻两行的相邻两个像素之间的所述新增行像素与相邻两列的相邻两个像素之间的所述新增列像素的中心的新增中心像素对应的灰度值；其中，所述相邻两行的相邻两个像素与所述相邻两列的相邻两个像素重合；
S4、在每行像素中的相邻两个像素之间插入所述新增行像素，在每列像素中的相邻两个像素之间插入所述新增列像素，以及在所述新增行像素和所述列像素的中心插入中心像素，以获取所述原始图像对应重构的超分辨率图像。
优选地，步骤S2中，所述根据每个所述行像素组中的像素之间的可视关系，获取每行像素中的相邻两个像素之间的新增行像素对应的灰度值具体包括：
S21、根据每个像素的灰度值，以及每个像素在所在像素行的行位置，确定每个像素在所在行像素组中的至少一个可视行像素；其中，每个像素与每个像素对应的至少一个所述可视行像素之间满足预设可视条件；
S22、根据每个像素的灰度值和行位置，以及每个像素对应的所 述可视行像素的灰度值和行位置，获取每行像素中的相邻两个像素之间的任意一个行位置的新增行像素对应的灰度值。
优选地，步骤S2中，所述根据每个所述列像素组中的像素之间的可视关系，获取每列像素中的相邻两个像素之间的新增列像素对应的灰度值，具体包括：
S21’、根据每个像素的灰度值，以及每个像素在所在像素列的列位置，确定每个像素在所在列像素组中的至少一个可视列像素；其中，每个像素与每个像素对应的至少一个所述可视列像素之间满足预设可视条件；
S22’、根据每个像素的灰度值和列位置，以及每个像素对应的所述可视列像素的灰度值和列位置，获取每列像素中的相邻两个像素之间的任意一个列位置的新增列像素对应的灰度值。
优选地，步骤S21中，所述预设可视条件包括：位于每个像素的行位置与每个像素对应的可视行像素的行位置之间的像素的灰度值均小于每个像素的灰度值和每个所述可视行像素的灰度值。
优选地，步骤S21’中，所述预设可视条件包括：位于每个像素的列位置与每个像素对应的可视列像素的列位置之间的像素的灰度值均小于每个像素的灰度值和每个所述可视列像素的灰度值。
优选地，步骤S22具体包括：
S221、根据每个像素的灰度值和行位置，以及每个像素对应的所述可视行像素的灰度值和行位置，分别建立每个像素与对应的每个所述可视行像素之间的可视函数关系；
S222、根据每个像素与对应的每个所述可视行像素之间的可视函数关系，选取与每行像素中的相邻两个像素相关联的可视函数关系；其中，与相邻两个像素相关联的可视函数关系为与相邻两个像素之间的行位置存在交集的可视函数关系；
S223、根据与相邻两个像素相关联的可视函数关系，获取相邻两个像素之间的所述任意一个行位置的新增行像素对应的灰度值。
优选地，步骤S22’具体包括：
S221’、根据每个像素的灰度值和列位置，以及每个像素对应的所述可视列像素的灰度值和列位置，分别建立每个像素与对应的每个所述可视列像素之间的可视函数关系；
S222’、根据每个像素与对应的每个所述可视列像素之间的可视函数关系，选取与每列像素中的相邻两个像素相关联的可视函数关系；其中，与相邻两个像素相关联的可视函数关系为与相邻两个像素之间的列位置存在交集的可视函数关系；
S223’、根据与相邻两个像素相关联的可视函数关系，获取相邻两个像素之间的所述任意一个列位置的新增列像素对应的灰度值。
优选地，步骤S223具体包括：
根据相邻两个像素之间的任意一个行位置，以及与相邻两个像素相关联的所有所述可视函数关系，获取所述任意一个行位置的新增行像素对应的多个初始灰度值；
将多个所述初始灰度值的平均值作为所述任意一个行位置的新增行像素对应的灰度值。
优选地，步骤S223’具体包括：
根据相邻两个像素之间的任意一个列位置，以及与相邻两个像素相关联的所有所述可视函数关系，获取所述任意一个列位置的新增列像素对应的多个初始灰度值；
将多个所述初始灰度值的平均值作为所述任意一个列位置的新增列像素对应的灰度值。
优选地，步骤S3具体包括：将所述相邻两行的相邻两个像素之间的所述新增行像素的灰度值，与相邻两列的相邻两个像素之间的所述新增列像素的灰度值的平均值，作为所述新增中心像素对应的灰度值。
优选地，每行像素的相邻两个像素之间插入的所述新增行像素的数量相同，和/或，每列像素的相邻两个像素之间插入的所述新增列像素的数量相同。
第二方面，本申请提供一种超分辨率图像重构装置，包括：
采集模块，用于采集待重构的原始图像，所述原始图像包括多个呈阵列分布的像素；其中，每行像素包括多个由连续像素组成的行像素组，每列像素包括多个由连续像素组成的列像素组；
新增行列像素模块，用于根据每个所述行像素组中的像素之间的可视关系，获取每行像素中的相邻两个像素之间的新增行像素对应的灰度值，以及，根据每个所述列像素组中的像素之间的可视关系，确定每列像素中的相邻两个像素之间的新增列像素对应的灰度值；
新增中心像素模块，用于根据所述新增行像素和所述新增列像素，确定在相邻两行的相邻两个像素之间的所述新增行像素与相邻两列的相邻两个像素之间的所述新增列像素的围绕区域中的新增中心像素对应的灰度值；其中，所述相邻两行的相邻两个像素与所述相邻两列的相邻两个像素重合；
插入模块，用于在每行像素中的相邻两个像素之间插入所述新增行像素，在每列像素中的相邻两个像素之间插入所述新增列像素，以及在所述新增行像素和所述列像素的围绕区域中插入中心像素，以获取所述原始图像对应重构的超分辨率图像。
优选地，所述新增行列像素模块具体包括：
行像素可视关系单元，用于根据每个像素的灰度值，以及每个像素在所在像素行的行位置，确定每个像素在所在行像素组中的至少一个可视行像素；其中，每个像素与每个像素对应的至少一个所述可视行像素之间满足预设可视条件；
新增行像素灰度值获取单元，用于根据每个像素的灰度值和行位置，以及每个像素对应的所述可视行像素的灰度值和行位置，获取每行像素中的相邻两个像素之间的任意一个行位置的新增行像素对应的灰度值。
优选地，所述新增行列像素模块具体包括：
列像素可视关系单元，用于根据每个像素的灰度值，以及每个像素在所在像素列的列位置，确定每个像素在所在列像素组中的至少一个可视列像素；其中，每个像素与每个像素对应的至少一个所述可视 列像素之间满足预设可视条件；
新增列像素灰度值获取单元，用于根据每个像素的灰度值和列位置，以及每个像素对应的所述可视列像素的灰度值和列位置，获取每列像素中的相邻两个像素之间的任意一个列位置的新增列像素对应的灰度值。
优选地，所述新增行像素灰度值获取单元具体用于：
根据每个像素的灰度值和行位置，以及每个像素对应的所述可视行像素的灰度值和行位置，分别建立每个像素与对应的每个所述可视行像素之间的可视函数关系；
根据每个像素与对应的每个所述可视行像素之间的可视函数关系，选取与每行像素中的相邻两个像素相关联的可视函数关系；其中，与相邻两个像素相关联的可视函数关系为与相邻两个像素之间的行位置存在交集的可视函数关系；
根据与相邻两个像素相关联的可视函数关系，获取相邻两个像素之间的所述任意一个行位置的新增行像素对应的灰度值。
优选地，所述新增列像素灰度值获取单元，具体用于：
根据每个像素的灰度值和列位置，以及每个像素对应的所述可视列像素的灰度值和列位置，分别建立每个像素与对应的每个所述可视列像素之间的可视函数关系；
根据每个像素与对应的每个所述可视列像素之间的可视函数关系，选取与每列像素中的相邻两个像素相关联的可视函数关系；其中，与相邻两个像素相关联的可视函数关系为与相邻两个像素之间的列位置存在交集的可视函数关系；
根据与相邻两个像素相关联的可视函数关系，获取相邻两个像素之间的所述任意一个列位置的新增列像素对应的灰度值。
优选地，所述新增行像素灰度值获取单元，具体用于根据与相邻两个像素相关联的可视函数关系，获取相邻两个像素之间的所述任意一个行位置的新增行像素对应的灰度值，进一步包括：
根据相邻两个像素之间的任意一个行位置，以及与相邻两个像素 相关联的所有所述可视函数关系，获取所述任意一个行位置的新增行像素对应的多个初始灰度值；
将多个所述初始灰度值的平均值作为所述任意一个行位置的新增行像素对应的灰度值。
优选地，所述新增列像素灰度值获取单元，具体用于根据与相邻两个像素相关联的可视函数关系，获取相邻两个像素之间的所述任意一个列位置的新增列像素对应的灰度值，进一步包括：
根据相邻两个像素之间的任意一个列位置，以及与相邻两个像素相关联的所有所述可视函数关系，获取所述任意一个列位置的新增列像素对应的多个初始灰度值；
将多个所述初始灰度值的平均值作为所述任意一个列位置的新增列像素对应的灰度值。
优选地，所述新增中心像素模块，具体用于：
选择重合的相邻两行的相邻两个像素和相邻两列的相邻两个像素；
将所述相邻两行的相邻两个像素之间的所述新增行像素的灰度值，与相邻两列的相邻两个像素之间的所述新增列像素的灰度值的平均值，作为所述新增中心像素对应的灰度值。
第三方面，本申请提供一种电子设备，包括存储器、处理器及存储在存储器上并可在处理器上运行的计算机程序，所述处理器执行所述程序时实现如上所述的超分辨率图像重构方法。
第四方面，本申请提供一种计算机可读存储介质，其上存储有计算机程序，该计算机程序被处理器执行时实现如上所述的超分辨率图像重构方法的步骤。
有益效果
本申请提供一种超分辨率图像重构方法及装置，根据原始图像中的每个像素与其所在行像素组或列像素组中的其他像素之间的可视关系，得知需要在原始图像的每行或每列的每相邻两个像素之间需要 插入的新增像素对应的灰度值，并将该灰度值的新增像素插入每行或每列的相邻两个像素中，从而得到将原始图像重构的超分辨率图像。由于该超分辨率图像重构方法充分考虑了原始图像的局部像素之间的内在联系，因此可以较好地保持图像的细节信息，锯齿、模糊、颗粒感等问题明显减弱，改善了原始图像经拉伸后的显示效果。此外，本发明方法原理简单、计算量小，实时处理性高，保障了信号高速传输、系统高效运行。
附图说明
图1为本申请实施例提供的超分辨率图像重构方法的流程示意图；
图2为本申请实施例提供的原始图像的任意一行像素的行位置与灰度值直方图；
图3为本申请实施例提供的原始图像的一个行像素组中的像素的行位置与灰度值直方图；
图4为本申请实施例提供的原始图像的另一个行像素组中的像素的行位置与灰度值直方图；
图5为本申请实施例提供的原始图像重构为超分辨率图像的过程示意图；
图6为本申请实施例提供的超分辨率图像重构装置的结构示意图；
图7为本申请实施例提供的电子设备的结构示意图。
本发明的实施方式
为使本申请的目的、技术方案及效果更加清楚、明确，以下参照附图并举实施例对本申请进一步详细说明。应当理解，此处所描述的具体实施例仅用以解释本申请，并不用于限定本申请。
图1为本申请实施例提供的超分辨率图像重构方法的流程示意图，如图1所示，本申请实施例提供一种超分辨率图像重构方法，包括：
S1、采集待重构的原始图像，原始图像包括多个呈阵列分布的像素；其中，每行像素包括多个由连续像素组成的行像素组，每列像素包括多个由连续像素组成的列像素组。
其中，连续是指每个像素组中的像素是每行或每列中顺序排列的像素点，即每个像素与邻近的行像素构成行像素组，每个像素与邻近的列像素构成列像素组。
具体地，为了充分考虑邻近像素之间的内在联系，将每行像素划分为多个行像素组，每个行像素组由第一预设数量的连续像素组成，同样地，将每列像素划分为多个列像素组，每个列像素组由第二预设数量的连续像素组成，以方便在后续步骤中在每个行像素组和每个列像素组中确定每个像素与其邻近像素之间的联系，以方便在将原始图像拉伸之后，在相邻两个像素之间填充最佳灰度值的新增行像素或新增列像素。
可以理解的是，实际上各个行像素组之间，各个列像素组之间，或者行像素组与列像素组之间，所包括的连续像素的数量可以相同，也可以不同，因此第一预设数量和第二预设数量并不是恒定值，且第一预设数量和第二预设数量也不一定相同，可以根据原始图像的重点重构部分和非重点重构部分设置第一预设数量和第二预设数量的具体数值，例如使重点重构部分的像素组包括较多的像素点，第一预设数量和第二预设数量就较大，比如6-10，而非重点重构部分的像素组包括较少的像素点，第一预设数量和第二预设数量就较小，比如1-5。另外，每个像素也不一定仅属于一个行像素组或一个列像素组，即相邻两个像素组可以包括重合的像素，只要预先确定每个像素组包括的具体像素点即可。
S2、根据每个行像素组中的像素之间的可视关系，获取每行像素中的相邻两个像素之间的新增行像素对应的灰度值，以及，根据每个列像素组中的像素之间的可视关系，确定每列像素中的相邻两个像素之间的新增列像素对应的灰度值。
具体地，位于同一行或同一列的像素之间存在可视关系或非可视 关系，在每个像素组中选取与每个像素具有可视关系的像素点的灰度值作为确定每相邻两个像素之间需要插入的新增像素的灰度值的依据，由此建立邻近像素之间的内在联系，从而根据每相邻两个像素与其在同一像素组的邻近像素之间的可视关系来确定这两个相邻两个像素之间需要插入的新增像素的灰度值。
S3、根据新增行像素和新增列像素，确定在相邻两行的相邻两个像素之间的新增行像素与相邻两列的相邻两个像素之间的新增列像素的围绕区域中的新增中心像素对应的灰度值；其中，相邻两行的相邻两个像素与相邻两列的相邻两个像素重合。
具体地，如图5所示，黑圆点为原始图像的像素，白圆点为新增行像素，矩形为新增列像素，三角形为新增中心像素，图5为将原始图像扩大2倍得到超分辨率图像时的示意图。在每行像素的相邻两个像素之间插入新增行像素，以及在每列像素的相邻两列像素之间插入新增列像素之后，对于相邻两行的相邻两个像素与相邻两列的相邻两个像素重合的四个像素(例如图5中的正方形虚线框)来说，这其中的新增行像素与新增列像素的围绕区域中存在空白像素点，因此需要在该围绕区域中插入新增中心像素。本实施例中的新增中心像素的灰度值由围绕其的两个新增行像素和新增列像素的灰度值决定。
S4、在每行像素中的相邻两个像素之间插入新增行像素，在每列像素中的相邻两个像素之间插入新增列像素，以及在新增行像素和列像素的围绕区域中插入中心像素，以获取原始图像对应重构的超分辨率图像。
具体地，在每行像素的相邻两个像素之间插入上述求得的灰度值的新增行像素，在每列像素的相邻两个像素之间插入上述求得的灰度值的新增列像素，并在相邻两行的相邻两个像素之间的新增行像素与相邻两列的相邻两个像素之间的新增列像素的围绕区域中插入新增中心像素，从而将原始图像重构为超分辨率图像。
本申请实施例提供的超分辨率图像重构方法，根据原始图像中的每个像素与其所在行像素组或列像素组中的其他像素之间的可视关 系，得知需要在原始图像的每行或每列的每相邻两个像素之间需要插入的新增行像素和新增列像素对应的灰度值，然后根据新增行像素和新增列像素确定重合的相邻两行的相邻两个像素和相邻两列的相邻两个像素的围绕区域的新增中心像素对应的灰度值，并将符合对应灰度值的新增像素插入每行或每列的相邻两个像素中，以及将符合对应灰度值的新增中心像素插入新增行像素和新增列像素的围绕区域中，从而得到将原始图像重构的超分辨率图像。由于该超分辨率图像重构方法充分考虑了原始图像的局部像素之间的内在联系，因此可以较好地保持图像的细节信息，锯齿、模糊、颗粒感等问题明显减弱，改善了原始图像经拉伸后的显示效果。此外，本发明方法原理简单、计算量小，实时处理性高，保障了信号高速传输、系统高效运行。
需要说明的是，由于每行像素的相邻两个像素之间填充新增行像素与在每列像素的相邻两个像素之间填充新增列像素的过程和步骤相同，以下实施例以在每行像素的相邻两个像素之间填充新增行像素为例，对该超分辨率图像重构方法进行进一步说明。
步骤S2中，根据每个行像素组中的像素之间的可视关系，获取每行像素中的相邻两个像素之间的新增行像素对应的灰度值具体包括：
S21、根据每个像素的灰度值，以及每个像素在所在像素行的行位置，确定每个像素在所在行像素组中的至少一个可视行像素。其中，每个像素与每个像素对应的至少一个可视行像素之间满足预设可视条件，预设可视条件包括：位于每个像素的行位置与每个像素对应的可视行像素的行位置之间的像素的灰度值均小于每个像素的灰度值和每个可视行像素的灰度值。
具体地，如图2所示，将像素的行位置作为横坐标(行位置的数值可自行设置，本实施例的行位置1、2、3……分别表示处于同一行的第1个、第2个、第3个……像素)，像素的灰度值作为纵坐标，绘制行位置-灰度值直方图。本申请实施例的可视关系判断准则为：若位于同一行的两个像素可视，则这两个像素之间的任意一个行位置满足以下公式：
Figure PCTCN2021107498-appb-000001
其中，(Xi，Pi)和(Xj，Pj)分别为位于同一行的两个像素的行位置值和灰度值，(Xm，Pm)为这两个像素之间的任意一个行位置的像素的行位置和灰度值。
上述公式的含义为：位于同一行的两个具有可视关系的像素之间的任意一个行位置的像素的灰度值均小于这两个像素的灰度值，也即，图2中的具有可视关系的像素的灰度值之间的连线没有被其他像素的灰度值直方条阻断，将此作为预设可视条件，反之，如果两个像素的灰度值之间的连线被这两个像素之间的其他像素的灰度值直方条阻断，则这两个像素之间不具有可视关系。
进一步地，根据预设可视条件，确定每个像素在所在行像素组中的至少一个可视行像素，例如：设第一预设数量为4，图3和图4分别为两个由4个连续像素(例如图5中的长方形虚线框)组成的行像素组的行位置-灰度值直方图，设这4个连续像素的行位置分别为行位置1、行位置2、行位置3和行位置4。
如图3所示，行位置1的像素分别与行位置2、行位置3和行位置4的像素具有可视关系，因此行位置1的像素在该行像素组中的可视像素为行位置2、行位置3和行位置4的像素；同理，行位置2的像素在该行像素组中的可视像素为行位置1、行位置3和行位置4的像素；行位置3的像素在该行像素组中的可视像素为行位置1、行位置2和行位置4的像素；行位置4的像素在该行像素组中的可视像素为行位置1、行位置2和行位置3的像素。
如图4所示，行位置1的像素在该行像素组中的可视像素为行位置2和行位置3的像素；同理，行位置2的像素在该行像素组中的可视像素为行位置1和行位置3的像素；行位置3的像素在该行像素组中的可视像素为行位置1、行位置2和行位置4的像素；行位置4的像素在该行像素组中的可视像素为行位置3的像素。
S22、根据每个像素的灰度值和行位置，以及每个像素对应的可 视行像素的灰度值和行位置，获取每行像素中的相邻两个像素之间的任意一个行位置的新增行像素对应的灰度值。
进一步地，步骤S22具体包括：
S221、根据每个像素的灰度值和行位置，以及每个像素对应的可视行像素的灰度值和行位置，分别建立每个像素与对应的每个可视行像素之间的可视函数关系。
例如，若图3中的行位置1、行位置2、行位置3和行位置4的像素的灰度值分别为240、160、140和210，根据行位置1的像素与行位置2的像素在直方图中的连线的斜率为：(160-240)/(2-1)＝-80，得到行位置1的像素与行位置2的像素之间的可视函数关系为：P12＝-80X12+320，其中，X12为行位置1和行位置2之间的任意一个行位置，P12为行位置1和行位置2之间的任意一个行位置的像素对应的灰度值。
同理，行位置1的像素与行位置3的像素之间的可视函数关系为：P13＝-50X13+290，其中，X13为行位置1与行位置3之间的任意一个行位置，P13为行位置1和行位置3之间的任意一个行位置的像素对应的灰度值。
行位置1的像素与行位置4的像素之间的可视函数关系为：P14＝-10X14+250，其中，X14为行位置1与行位置3之间的任意一个行位置，P14为行位置1和行位置4之间的任意一个行位置的像素对应的灰度值。
行位置2的像素与行位置3的像素之间的可视函数关系为：P23＝-20X23+200，其中，X23为行位置2与行位置3之间的任意一个行位置，P23为行位置2和行位置3之间的任意一个行位置的像素对应的灰度值。
行位置2的像素与行位置4的像素之间的可视函数关系为：P24＝25X24+110，其中，X24为行位置2与行位置4之间的任意一个行位置，P24为行位置2和行位置4之间的任意一个行位置的像素对应的灰度值。
行位置3的像素与行位置4的像素之间的可视函数关系为：P34＝70X34-70，其中，X24为行位置2与行位置4之间的任意一个行位置。
同样地，若图4中的行位置1、行位置2、行位置3和行位置4的像素的灰度值分别为240、160、200和120，则行位置1的像素与行位置2的像素之间的可视函数关系为：P12＝-80X12+320，行位置1的像素与行位置3的像素之间的可视函数关系为：P13＝-20X13+260；行位置1的像素与行位置4的像素之间的可视函数关系为：P14＝-40X14+280。行位置2的像素与行位置3的像素之间的可视函数关系为：P23＝40X23+80。
S222、根据每个像素与对应的每个可视行像素之间的可视函数关系，选取与每行像素中的相邻两个像素相关联的可视函数关系；其中，与相邻两个像素相关联的可视函数关系为与相邻两个像素之间的行位置存在交集的可视函数关系。
具体地，与相邻两个像素之间的行位置存在交集的可视函数关系，是指在直方图中，这两个相邻的像素分别与对应的可视像素的连线，需要经过这两个相邻两个像素之间，才会将这些连线对应的可视函数关系作为与这两个相邻的像素相关联的可视函数关系。
例如，在行位置1的像素和行位置2的像素之间插入新增行像素时，在图3中应当选择行位置1的像素与行位置2的像素之间的可视函数关系P12，行位置1的像素与行位置3的像素之间的可视函数关系P13，以及行位置1的像素与行位置4的像素之间的可视函数关系P14，作为与行位置1的像素和行位置2的像素相关联的可视函数关系，而在图4中应当选择行位置1的像素与行位置2的像素之间的可视函数关系P12，以及行位置1的像素与行位置3的像素之间的可视函数关系P13，作为与行位置1的像素和行位置2的像素相关联的可视函数关系。
在行位置2的像素和行位置3的像素之间插入新增行像素时，在图3中应当选择行位置2的像素与行位置3的像素之间的可视函数关系P23，行位置2的像素与行位置4的像素之间的可视函数关系P24，以及 行位置3的像素与行位置1的像素之间的可视函数关系P31(即为行位置1的像素与行位置3的像素之间的可视函数关系P13)，作为与行位置2和行位置3相关联的可视函数关系，而在图4中应当选择行位置2的像素与行位置3的像素之间的可视函数关系P23，以及行位置3的像素与行位置1的像素之间的可视函数关系P31，作为与行位置2和行位置3相关联的可视函数关系。
S223、根据与相邻两个像素相关联的可视函数关系，获取相邻两个像素之间的任意一个行位置的新增行像素对应的灰度值。
进一步地，步骤S223具体包括：根据相邻两个像素之间的任意一个行位置，以及与相邻两个像素相关联的所有可视函数关系，获取任意一个行位置的新增行像素对应的多个初始灰度值，并将多个初始灰度值的平均值作为任意一个行位置的新增行像素对应的灰度值。
例如，在行位置1的像素和行位置2的像素之间的任意一个行位置x插入新增行像素时，图3中，Px＝(P12+P13+P14)/3，而图4中，Px＝(P12+P13)/2，若选择在行位置x＝1.5处插入新增行像素，则图3中由x＝1.5，得到P12＝200，P13＝215，P14＝235，因此P1.5＝(200+215+235)/3＝216，从而在1.5的位置插入灰度值为216的像素；而在图4中，由x＝1.5，得到P12＝200，P13＝230，因此P1.5＝(200+230)/2＝215，从而在1.5的位置插入灰度值为215的像素。
在行位置2的像素和行位置3的像素之间的任意一个行位置x插入新增行像素时，图3中，Px＝(P23+P24+P31)/3，而图4中，Px＝(P12+P13)/2，若选择在行位置x＝2.5处插入新增行像素，则图3中由x＝2.5，得到P23＝150，P24＝172.5，P31＝165，因此P2.5＝(150+172.5+165)/3＝162.5，从而在2.5的位置插入灰度值为162.5的像素；而在图4中，由x＝2.5，得到P12＝120，P13＝210，因此P2.5＝(120+210)/2＝165，从而在2.5的位置插入灰度值为165的像素。
进一步地，步骤S3具体包括：
S31、选择重合的相邻两行的相邻两个像素和相邻两列的相邻两个像素。
S32、将相邻两行的相邻两个像素之间的新增行像素的灰度值，与相邻两列的相邻两个像素之间的新增列像素的灰度值的平均值，作为新增中心像素对应的灰度值。
具体地，选择重合的相邻两行的相邻两个像素和相邻两列的相邻两个像素(例如图5中的虚线框中的4个黑色圆点)，插入新增行像素(例如图5中的虚线框中的2个白色圆点)和新增列像素(例如图5中的虚线框中的2个矩形)，然后求相邻两行的相邻两个像素之间的新增行像素例如图5中的虚线框中的2个黑色圆点)的灰度值，与相邻两列的相邻两个像素之间的新增列像素的灰度值例如图5中的虚线框中的2个白色圆点)的平均值，将该平均值作为新增中心像素(例如图5中的虚线框中的三角形)对应的灰度值。
需要说明的是，为了保证原始图像经过扩大后得到的重构的超分辨率图像的显示均一性，每行像素的相邻两个像素之间插入的新增行像素的数量相同，每列像素的相邻两个像素之间插入的新增列像素的数量相同，例如在将4K分辨率的原始图像扩大为8K超分辨率图像时，在每行的每相邻两个像素之间均插入一个新增行像素，且在每列的每相邻两个像素之间均插入一个新增列像素，而不是在部分相邻两个像素之间不插入像素，而在部分相邻两个像素之间插入多于一个新增像素，即，在每相邻两个像素之间均匀插入同样数量的像素，使得最终得到的超分辨率图像的显示效果均一，不会出现部分区域模糊、锯齿和颗粒感强，而部分区域非常清晰的不均一现象。
可以理解的是，为了进一步保证最终获得的超分辨率图像的显示均一性，每个新增像素的行位置与相邻两个像素的行位置之间的间距还可以保持一致，例如均在相邻两个像素的中点插入新增像素。
基于上述实施例，如图6所示，本申请实施例还提供一种超分辨率图像重构装置，包括：
采集模块601，用于采集待重构的原始图像，所述原始图像包括多个呈阵列分布的像素；其中，每行像素包括多个由连续像素组成的行像素组，每列像素包括多个由连续像素组成的列像素组；
新增行列像素模块602，用于根据每个所述行像素组中的像素之间的可视关系，获取每行像素中的相邻两个像素之间的新增行像素对应的灰度值，以及，根据每个所述列像素组中的像素之间的可视关系，确定每列像素中的相邻两个像素之间的新增列像素对应的灰度值；
新增中心像素模块603，用于根据所述新增行像素和所述新增列像素，确定在相邻两行的相邻两个像素之间的所述新增行像素与相邻两列的相邻两个像素之间的所述新增列像素的围绕区域中的新增中心像素对应的灰度值；其中，所述相邻两行的相邻两个像素与所述相邻两列的相邻两个像素重合；
插入模块604，用于在每行像素中的相邻两个像素之间插入所述新增行像素，在每列像素中的相邻两个像素之间插入所述新增列像素，以及在所述新增行像素和所述列像素的围绕区域中插入中心像素，以获取所述原始图像对应重构的超分辨率图像。
本申请实施例提供的超分辨率图像重构装置，根据原始图像中的每个像素与其所在行像素组或列像素组中的其他像素之间的可视关系，得知需要在原始图像的每行或每列的每相邻两个像素之间需要插入的新增像素对应的灰度值，并将该灰度值的新增像素插入每行或每列的相邻两个像素中，从而得到将原始图像重构的超分辨率图像。由于该超分辨率图像重构方法充分考虑了原始图像的局部像素之间的内在联系，因此可以较好地保持图像的细节信息，锯齿、模糊、颗粒感等问题明显减弱，改善了原始图像经拉伸后的显示效果。此外，本发明方法原理简单、计算量小，实时处理性高，保障了信号高速传输、系统高效运行。
图7为本发明实施例提供的电子设备的结构示意图，如图7所示，本发明实施例提供一种电子设备，该电子设备可以包括：处理器(processor)701、通信接口(Communications Interface)702、存储器(memory)703和通信总线704，其中，处理器701，通信接口702，存储器703通过通信总线704完成相互间的通信。通信接口702可以用于服务器与智能电视之间的信息传输。处理器701可以调用存储器703中的 逻辑指令，以执行如下方法，例如包括：S1、采集待重构的原始图像，原始图像包括多个呈阵列分布的像素；其中，每行像素包括多个由连续像素组成的行像素组，每列像素包括多个由连续像素组成的列像素组；S2、根据每个行像素组中的像素之间的可视关系，获取每行像素中的相邻两个像素之间的新增行像素对应的灰度值，以及，根据每个列像素组中的像素之间的可视关系，确定每列像素中的相邻两个像素之间的新增列像素对应的灰度值；S3、根据新增行像素和新增列像素，确定在相邻两行的相邻两个像素之间的新增行像素与相邻两列的相邻两个像素之间的新增列像素的中心的新增中心像素对应的灰度值；其中，相邻两行的相邻两个像素与相邻两列的相邻两个像素重合；S4、在每行像素中的相邻两个像素之间插入新增行像素，在每列像素中的相邻两个像素之间插入新增列像素，以及在新增行像素和列像素的中心插入中心像素，以获取所述原始图像对应重构的超分辨率图像。
本实施例还提供一种非暂态计算机可读存储介质，所述非暂态计算机可读存储介质存储计算机指令，所述计算机指令使所述计算机执行上述各方法实施例所提供的方法，例如包括：S1、采集待重构的原始图像，原始图像包括多个呈阵列分布的像素；其中，每行像素包括多个由连续像素组成的行像素组，每列像素包括多个由连续像素组成的列像素组；S2、根据每个行像素组中的像素之间的可视关系，获取每行像素中的相邻两个像素之间的新增行像素对应的灰度值，以及，根据每个列像素组中的像素之间的可视关系，确定每列像素中的相邻两个像素之间的新增列像素对应的灰度值；S3、根据新增行像素和新增列像素，确定在相邻两行的相邻两个像素之间的新增行像素与相邻两列的相邻两个像素之间的新增列像素的中心的新增中心像素对应的灰度值；其中，相邻两行的相邻两个像素与相邻两列的相邻两个像素重合；S4、在每行像素中的相邻两个像素之间插入新增行像素，在每列像素中的相邻两个像素之间插入新增列像素，以及在新增行像素和列像素的中心插入中心像素，以获取所述原始图像对应重构的超分辨率图像。
可以理解的是，对本领域普通技术人员来说，可以根据本申请的技术方案及其发明构思加以等同替换或改变，而所有这些改变或替换都应属于本申请所附的权利要求的保护范围。

Claims (19)

  1. 一种超分辨率图像重构方法，其包括：
    S1、采集待重构的原始图像，所述原始图像包括多个呈阵列分布的像素；其中，每行像素包括多个由连续像素组成的行像素组，每列像素包括多个由连续像素组成的列像素组；
    S2、根据每个所述行像素组中的像素之间的可视关系，获取每行像素中的相邻两个像素之间的新增行像素对应的灰度值，以及，根据每个所述列像素组中的像素之间的可视关系，确定每列像素中的相邻两个像素之间的新增列像素对应的灰度值；
    S3、根据所述新增行像素和所述新增列像素，确定在相邻两行的相邻两个像素之间的所述新增行像素与相邻两列的相邻两个像素之间的所述新增列像素的围绕区域中的新增中心像素对应的灰度值；其中，所述相邻两行的相邻两个像素与所述相邻两列的相邻两个像素重合；
    S4、在每行像素中的相邻两个像素之间插入所述新增行像素，在每列像素中的相邻两个像素之间插入所述新增列像素，以及在所述新增行像素和所述列像素的围绕区域中插入中心像素，以获取所述原始图像对应重构的超分辨率图像。
  2. 如权利要求1所述的超分辨率图像重构方法，其中，步骤S2中，所述根据每个所述行像素组中的像素之间的可视关系，获取每行像素中的相邻两个像素之间的新增行像素对应的灰度值，具体包括：
    S21、根据每个像素的灰度值，以及每个像素在所在像素行的行位置，确定每个像素在所在行像素组中的至少一个可视行像素；其中，每个像素与每个像素对应的至少一个所述可视行像素之间满足预设可视条件；
    S22、根据每个像素的灰度值和行位置，以及每个像素对应的所述可视行像素的灰度值和行位置，获取每行像素中的相邻两个像素之间的任意一个行位置的新增行像素对应的灰度值。
  3. 如权利要求1所述的超分辨率图像重构方法，其中，步骤S2中，所述根据每个所述列像素组中的像素之间的可视关系，获取每列像素中的相邻两个像素之间的新增列像素对应的灰度值，具体包括：
    S21’、根据每个像素的灰度值，以及每个像素在所在像素列的列位置，确定每个像素在所在列像素组中的至少一个可视列像素；其中，每个像素与每个像素对应的至少一个所述可视列像素之间满足预设可视条件；
    S22’、根据每个像素的灰度值和列位置，以及每个像素对应的所述可视列像素的灰度值和列位置，获取每列像素中的相邻两个像素之间的任意一个列位置的新增列像素对应的灰度值。
  4. 如权利要求2所述的超分辨率图像重构方法，其中，步骤S21中，所述预设可视条件包括：位于每个像素的行位置与每个像素对应的可视行像素的行位置之间的像素的灰度值均小于每个像素的灰度值和每个所述可视行像素的灰度值。
  5. 如权利要求3所述的超分辨率图像重构方法，其中，步骤S21’中，所述预设可视条件包括：位于每个像素的列位置与每个像素对应的可视列像素的列位置之间的像素的灰度值均小于每个像素的灰度值和每个所述可视列像素的灰度值。
  6. 如权利要求2所述的超分辨率图像重构方法，其中，步骤S22具体包括：
    S221、根据每个像素的灰度值和行位置，以及每个像素对应的所述可视行像素的灰度值和行位置，分别建立每个像素与对应的每个所述可视行像素之间的可视函数关系；
    S222、根据每个像素与对应的每个所述可视行像素之间的可视函数关系，选取与每行像素中的相邻两个像素相关联的可视函数关系；其中，与相邻两个像素相关联的可视函数关系为与相邻两个像素之间的行位置存在交集的可视函数关系；
    S223、根据与相邻两个像素相关联的可视函数关系，获取相邻两个像素之间的所述任意一个行位置的新增行像素对应的灰度值。
  7. 如权利要求3所述的超分辨率图像重构方法，其中，步骤S22’具体包括：
    S221’、根据每个像素的灰度值和列位置，以及每个像素对应的所述可视列像素的灰度值和列位置，分别建立每个像素与对应的每个所述可视列像素之间的可视函数关系；
    S222’、根据每个像素与对应的每个所述可视列像素之间的可视函数关系，选取与每列像素中的相邻两个像素相关联的可视函数关系；其中，与相邻两个像素相关联的可视函数关系为与相邻两个像素之间的列位置存在交集的可视函数关系；
    S223’、根据与相邻两个像素相关联的可视函数关系，获取相邻两个像素之间的所述任意一个列位置的新增列像素对应的灰度值。
  8. 如权利要求6所述的超分辨率图像重构方法，其中，步骤S223具体包括：
    根据相邻两个像素之间的任意一个行位置，以及与相邻两个像素相关联的所有所述可视函数关系，获取所述任意一个行位置的新增行像素对应的多个初始灰度值；
    将多个所述初始灰度值的平均值作为所述任意一个行位置的新增行像素对应的灰度值。
  9. 如权利要求7所述的超分辨率图像重构方法，其中，步骤S223’具体包括：
    根据相邻两个像素之间的任意一个列位置，以及与相邻两个像素相关联的所有所述可视函数关系，获取所述任意一个列位置的新增列像素对应的多个初始灰度值；
    将多个所述初始灰度值的平均值作为所述任意一个列位置的新增列像素对应的灰度值。
  10. 如权利要求1所述的超分辨率图像重构方法，其中，步骤S3具体包括：
    S31、选择重合的相邻两行的相邻两个像素和相邻两列的相邻两个像素；
    S32、将所述相邻两行的相邻两个像素之间的所述新增行像素的灰度值，与相邻两列的相邻两个像素之间的所述新增列像素的灰度值的平均值，作为所述新增中心像素对应的灰度值。
  11. 如权利要求1所述的超分辨率图像重构方法，其中，每行像素的相邻两个像素之间插入的所述新增行像素的数量相同，和/或，每列像素的相邻两个像素之间插入的所述新增列像素的数量相同。
  12. 一种超分辨率图像重构装置，其包括：
    采集模块，用于采集待重构的原始图像，所述原始图像包括多个呈阵列分布的像素；其中，每行像素包括多个由连续像素组成的行像素组，每列像素包括多个由连续像素组成的列像素组；
    新增行列像素模块，用于根据每个所述行像素组中的像素之间的可视关系，获取每行像素中的相邻两个像素之间的新增行像素对应的灰度值，以及，根据每个所述列像素组中的像素之间的可视关系，确定每列像素中的相邻两个像素之间的新增列像素对应的灰度值；
    新增中心像素模块，用于根据所述新增行像素和所述新增列像素，确定在相邻两行的相邻两个像素之间的所述新增行像素与相邻两列的相邻两个像素之间的所述新增列像素的围绕区域中的新增中心像素对应的灰度值；其中，所述相邻两行的相邻两个像素与所述相邻两列的相邻两个像素重合；
    插入模块，用于在每行像素中的相邻两个像素之间插入所述新增行像素，在每列像素中的相邻两个像素之间插入所述新增列像素，以及在所述新增行像素和所述列像素的围绕区域中插入中心像素，以获取所述原始图像对应重构的超分辨率图像。
  13. 如权利要求12所述的超分辨率图像重构装置，其中，所述新增行列像素模块具体包括：
    行像素可视关系单元，用于根据每个像素的灰度值，以及每个像素在所在像素行的行位置，确定每个像素在所在行像素组中的至少一个可视行像素；其中，每个像素与每个像素对应的至少一个所述可视行像素之间满足预设可视条件；
    新增行像素灰度值获取单元，用于根据每个像素的灰度值和行位置，以及每个像素对应的所述可视行像素的灰度值和行位置，获取每行像素中的相邻两个像素之间的任意一个行位置的新增行像素对应的灰度值。
  14. 如权利要求12所述的超分辨率图像重构装置，其中，所述新增行列像素模块具体包括：
    列像素可视关系单元，用于根据每个像素的灰度值，以及每个像素在所在像素列的列位置，确定每个像素在所在列像素组中的至少一个可视列像素；其中，每个像素与每个像素对应的至少一个所述可视列像素之间满足预设可视条件；
    新增列像素灰度值获取单元，用于根据每个像素的灰度值和列位置，以及每个像素对应的所述可视列像素的灰度值和列位置，获取每列像素中的相邻两个像素之间的任意一个列位置的新增列像素对应的灰度值。
  15. 如权利要求13所述的超分辨率图像重构装置，其中，所述新增行像素灰度值获取单元具体用于：
    根据每个像素的灰度值和行位置，以及每个像素对应的所述可视行像素的灰度值和行位置，分别建立每个像素与对应的每个所述可视行像素之间的可视函数关系；
    根据每个像素与对应的每个所述可视行像素之间的可视函数关系，选取与每行像素中的相邻两个像素相关联的可视函数关系；其中，与相邻两个像素相关联的可视函数关系为与相邻两个像素之间的行位置存在交集的可视函数关系；
    根据与相邻两个像素相关联的可视函数关系，获取相邻两个像素之间的所述任意一个行位置的新增行像素对应的灰度值。
  16. 如权利要求14所述的超分辨率图像重构装置，其中，所述新增列像素灰度值获取单元，具体用于：
    根据每个像素的灰度值和列位置，以及每个像素对应的所述可视列像素的灰度值和列位置，分别建立每个像素与对应的每个所述可视 列像素之间的可视函数关系；
    根据每个像素与对应的每个所述可视列像素之间的可视函数关系，选取与每列像素中的相邻两个像素相关联的可视函数关系；其中，与相邻两个像素相关联的可视函数关系为与相邻两个像素之间的列位置存在交集的可视函数关系；
    根据与相邻两个像素相关联的可视函数关系，获取相邻两个像素之间的所述任意一个列位置的新增列像素对应的灰度值。
  17. 如权利要求15所述的超分辨率图像重构装置，其中，所述新增行像素灰度值获取单元，具体用于根据与相邻两个像素相关联的可视函数关系，获取相邻两个像素之间的所述任意一个行位置的新增行像素对应的灰度值，进一步包括：
    根据相邻两个像素之间的任意一个行位置，以及与相邻两个像素相关联的所有所述可视函数关系，获取所述任意一个行位置的新增行像素对应的多个初始灰度值；
    将多个所述初始灰度值的平均值作为所述任意一个行位置的新增行像素对应的灰度值。
  18. 如权利要求16所述的超分辨率图像重构装置，其中，所述新增列像素灰度值获取单元，具体用于根据与相邻两个像素相关联的可视函数关系，获取相邻两个像素之间的所述任意一个列位置的新增列像素对应的灰度值，进一步包括：
    根据相邻两个像素之间的任意一个列位置，以及与相邻两个像素相关联的所有所述可视函数关系，获取所述任意一个列位置的新增列像素对应的多个初始灰度值；
    将多个所述初始灰度值的平均值作为所述任意一个列位置的新增列像素对应的灰度值。
  19. 如权利要求12所述的超分辨率图像重构装置，其中，所述新增中心像素模块，具体用于：
    选择重合的相邻两行的相邻两个像素和相邻两列的相邻两个像素；
    将所述相邻两行的相邻两个像素之间的所述新增行像素的灰度值，与相邻两列的相邻两个像素之间的所述新增列像素的灰度值的平均值，作为所述新增中心像素对应的灰度值。
PCT/CN2021/107498 2021-06-24 2021-07-21 超分辨率图像重构方法及装置 Ceased WO2022267136A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US17/429,950 US12462338B2 (en) 2021-06-24 2021-07-21 Super-resolution image reconstruction method and device

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110702940.5 2021-06-24
CN202110702940.5A CN113487477B (zh) 2021-06-24 2021-06-24 超分辨率图像重构方法及装置

Publications (1)

Publication Number Publication Date
WO2022267136A1 true WO2022267136A1 (zh) 2022-12-29

Family

ID=77936036

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/107498 Ceased WO2022267136A1 (zh) 2021-06-24 2021-07-21 超分辨率图像重构方法及装置

Country Status (3)

Country Link
US (1) US12462338B2 (zh)
CN (1) CN113487477B (zh)
WO (1) WO2022267136A1 (zh)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113487477B (zh) * 2021-06-24 2023-12-01 Tcl华星光电技术有限公司 超分辨率图像重构方法及装置
CN115311146B (zh) * 2022-08-25 2026-04-14 深圳市洲明科技股份有限公司 图像数据处理方法、装置、计算机设备和存储介质

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0574245A2 (en) * 1992-06-11 1993-12-15 International Business Machines Corporation Method and apparatus for variable expansion and variable shrinkage of an image
WO2000041386A2 (en) * 1999-01-08 2000-07-13 Macronix International Co., Ltd Method and apparatus for scaling an image with locally selected interpolation functions
CN1543221A (zh) * 2003-11-04 2004-11-03 浙江大学 1/4像素精度插值的方法和装置
CN101226634A (zh) * 2008-01-30 2008-07-23 哈尔滨工业大学 基于核回归的单帧图像快速超分辨率重建方法
CN103810675A (zh) * 2013-09-09 2014-05-21 深圳市华星光电技术有限公司 图像超分辨率重构系统及方法
CN112508793A (zh) * 2020-12-22 2021-03-16 深圳开立生物医疗科技股份有限公司 一种图像缩放方法、装置及电子设备和存储介质

Family Cites Families (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5329614A (en) * 1991-02-07 1994-07-12 Unisys Corporation Method and apparatus for enlarging gray scale images
TW377431B (en) * 1995-04-14 1999-12-21 Hitachi Ltd Method and apparatus for changing resolution
JP4355744B2 (ja) * 2007-12-17 2009-11-04 シャープ株式会社 画像処理装置
JP5660300B2 (ja) * 2010-10-07 2015-01-28 セイコーエプソン株式会社 画像処理装置
JP2012231378A (ja) * 2011-04-27 2012-11-22 Olympus Corp 撮像装置及び画像生成方法
CN103810678B (zh) * 2014-01-21 2017-01-25 公安部第一研究所 一种背散射x射线人体安全检查设备畸变图像矫正方法
CN104574277A (zh) * 2015-01-30 2015-04-29 京东方科技集团股份有限公司 图像插值方法和图像插值装置
CN104599626B (zh) * 2015-03-02 2017-03-01 京东方科技集团股份有限公司 显示驱动方法和装置、采样区的生成方法和装置
CN104835440B (zh) * 2015-05-22 2017-08-29 京东方科技集团股份有限公司 图像的显示方法、显示装置
CN108960037B (zh) * 2018-04-28 2021-08-06 天津大学 基于邻居可视长度熵的不同生理状态脑电信号识别方法
CN112053284A (zh) * 2020-08-24 2020-12-08 海南太美航空股份有限公司 一种插入空像素点的图像处理方法、系统和电子设备
CN113487477B (zh) * 2021-06-24 2023-12-01 Tcl华星光电技术有限公司 超分辨率图像重构方法及装置

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0574245A2 (en) * 1992-06-11 1993-12-15 International Business Machines Corporation Method and apparatus for variable expansion and variable shrinkage of an image
WO2000041386A2 (en) * 1999-01-08 2000-07-13 Macronix International Co., Ltd Method and apparatus for scaling an image with locally selected interpolation functions
CN1543221A (zh) * 2003-11-04 2004-11-03 浙江大学 1/4像素精度插值的方法和装置
CN101226634A (zh) * 2008-01-30 2008-07-23 哈尔滨工业大学 基于核回归的单帧图像快速超分辨率重建方法
CN103810675A (zh) * 2013-09-09 2014-05-21 深圳市华星光电技术有限公司 图像超分辨率重构系统及方法
CN112508793A (zh) * 2020-12-22 2021-03-16 深圳开立生物医疗科技股份有限公司 一种图像缩放方法、装置及电子设备和存储介质

Also Published As

Publication number Publication date
US12462338B2 (en) 2025-11-04
US20240046414A1 (en) 2024-02-08
CN113487477B (zh) 2023-12-01
CN113487477A (zh) 2021-10-08

Similar Documents

Publication Publication Date Title
CN113596573B (zh) 图像处理方法、装置、电子设备和计算机可读存储介质
CN108287678B (zh) 一种基于虚拟现实的图像处理方法、装置、设备和介质
CN108109109B (zh) 一种超分辨率图像重构方法、装置、介质及计算设备
CN112399120A (zh) 电子装置及其控制方法
CN103531176B (zh) 一种Android终端的字体显示优化方法及系统
JP2020531931A5 (zh)
US11580621B2 (en) AI frame engine for mobile edge
WO2022267136A1 (zh) 超分辨率图像重构方法及装置
WO2019153731A1 (zh) 图像处理方法及系统
TW202141964A (zh) 人工智慧超解析度和接續尺寸調整電路的圖像處理裝置
CN108242063B (zh) 一种基于gpu加速的光场图像深度估计方法
CN113096045B (zh) 一种位图拉伸方法、位图拉伸装置及电子设备
CN107220934A (zh) 图像重建方法及装置
CN113436087B (zh) 一种图像二值化及显示方法、存储介质和终端设备
CN113316018B (zh) 时间信息在视频画面显示的叠加方法、设备及存储介质
CN109712578B (zh) 显示方法和显示装置
CN119067863A (zh) 一种同步式特征融合的水下图像增强方法及系统
CN115205752A (zh) 一种基于智能显示的液晶拼接lcd方法及系统
CN118037553A (zh) 基于2d-vmd和插值的图像增强方法
CN105468312B (zh) 一种图像显示方法及装置
CN111724303B (zh) 一种具有图像类型适应性的超分辨率图像处理方法与系统
CN115689850A (zh) 图像处理方法、显示装置及存储介质
US20150310595A1 (en) Local contrast enhancement method and apparatus
CN119906810B (zh) 一种高分辨率裸眼3d视频生成中的实时空洞填充方法
CN116309109B (zh) 一种基于亮度直方图均衡化的色阶映射方法

Legal Events

Date Code Title Description
WWE Wipo information: entry into national phase

Ref document number: 17429950

Country of ref document: US

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21946624

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21946624

Country of ref document: EP

Kind code of ref document: A1

WWG Wipo information: grant in national office

Ref document number: 17429950

Country of ref document: US