WO2016154966A1 - Method and system for image scaling based on edge self-adaptation - Google Patents

Method and system for image scaling based on edge self-adaptation Download PDF

Info

Publication number
WO2016154966A1
WO2016154966A1 PCT/CN2015/075696 CN2015075696W WO2016154966A1 WO 2016154966 A1 WO2016154966 A1 WO 2016154966A1 CN 2015075696 W CN2015075696 W CN 2015075696W WO 2016154966 A1 WO2016154966 A1 WO 2016154966A1
Authority
WO
WIPO (PCT)
Prior art keywords
edge
pixel
interpolated
interpolation
probability
Prior art date
Application number
PCT/CN2015/075696
Other languages
French (fr)
Chinese (zh)
Inventor
韩睿
汤仁君
郭若杉
罗杨
颜奉丽
汤晓莉
Original Assignee
中国科学院自动化研究所
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 中国科学院自动化研究所 filed Critical 中国科学院自动化研究所
Priority to PCT/CN2015/075696 priority Critical patent/WO2016154966A1/en
Publication of WO2016154966A1 publication Critical patent/WO2016154966A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00—Geometric image transformation in the plane of the image
    • G06T3/40—Scaling the whole image or part thereof

Definitions

  • the invention belongs to the field of image processing, and in particular relates to an image scaling method and system based on edge adaptation.
  • the original video source such as standard definition (720*576), high definition (1920*1080)
  • standard definition 720*576
  • high definition (1920*1080)
  • video scaling the performance requirements of the video scaling algorithm are very high.
  • interpolation-based video scaling methods There are two main types of traditional interpolation-based video scaling methods: low-pass filtering based interpolation and edge-based interpolation. Interpolation methods based on low-pass filtering, such as bilinear interpolation, bicubic interpolation, multi-phase interpolation, etc., while obtaining a smoother interpolated image, will cause high-frequency information loss in the image, blurring and aliasing at the edge of the image. . Based on the edge-based image interpolation method, the edge direction of the interpolation point is calculated by edge detection, and the interpolation points are interpolated along the edge direction to obtain a smooth image edge to avoid aliasing, but it will cause blurring in the texture area.
  • edge-based image interpolation method Based on the edge-based image interpolation method, the edge direction of the interpolation point is calculated by edge detection, and the interpolation points are interpolated along the edge direction to obtain a smooth image edge to avoid aliasing, but it will
  • the invention provides an image scaling method and system based on edge adaptation, so as to achieve the purpose of improving the definition of the interpolated image and reducing edge aliasing.
  • a first aspect of the present invention provides an image scaling method based on edge adaptation, including:
  • interpolation is performed according to the edge direction by using an edge interpolation method
  • the interpolation is performed according to the non-edge interpolation method
  • a second aspect of the present invention provides an edge adaptive image scaling system, including:
  • a coordinate calculation module configured to determine a position of the pixel to be interpolated in the original image
  • a strong edge detection module for calculating an edge direction and a strong edge probability of the pixel to be interpolated
  • An edge interpolation module configured to perform interpolation by using an edge interpolation method according to the edge direction when a strong edge probability of the pixel to be interpolated is not zero;
  • a non-edge interpolation module configured to perform interpolation according to a non-edge interpolation method when a strong edge probability of the pixel to be interpolated is zero;
  • a fusion module configured to use the strong edge probability and the complementary probability of the strong edge probability as edge interpolation fusion weights and non-edge interpolation fusion weights respectively, and the interpolation results obtained by the edge interpolation method and the non-edge interpolation method A weighted fusion is performed to obtain the final interpolated image.
  • the edge adaptive image scaling method of the present invention can use a large number of original points to perform interpolation processing with an arbitrary integer or non-integer integer, and perform interpolation processing in an arbitrary edge direction, so that the enlarged image edges are clear and the sawtooth phenomenon is avoided.
  • Embodiment 1 is a flowchart of Embodiment 1 of an image adaptive method based on edge adaptation according to the present invention
  • FIG. 2 is a schematic diagram of a Sobel gradient method according to Embodiment 1 of an edge-adaptive image scaling method according to the present invention
  • FIG. 3 is a schematic diagram showing the relationship between edge directions and setting parameters in the first embodiment of the image scaling method based on edge adaptation according to the present invention
  • Embodiment 4 is an incremental function relationship diagram constructed in Embodiment 1 of an edge adaptive image scaling method according to the present invention.
  • FIG. 5 is a functional relationship diagram between edge reliability and a constructor in the first embodiment of the image scaling method based on edge adaptation according to the present invention
  • FIG. 6 is an edge direction of the first embodiment of the image scaling method based on edge adaptation according to the present invention; a functional relationship diagram of consistency and slope variance;
  • FIG. 7 is a schematic diagram of a line intersection method according to Embodiment 1 of an edge adaptive image scaling method according to the present invention.
  • FIG. 8 is a structural block diagram of Embodiment 1 of an image scaling system based on edge adaptation according to the present invention.
  • FIG. 1 is a flowchart of Embodiment 1 of an edge-adaptive image scaling method according to the present invention.
  • the edge adaptive image scaling method of the present invention includes:
  • the input is the original image, that is, the low-resolution image I L
  • the image width and height are W L and H L
  • the interpolated high-resolution image is I H
  • the image width and height are W H and H H , respectively.
  • the coordinate in the low-resolution image can be obtained by the coordinate calculation unit (i L , j L ):
  • i L and j L respectively represent row coordinates and column coordinates of the position of the pixel to be interpolated in the original image, that is, the low-resolution image
  • i H and j H respectively represent the pixel to be interpolated in the image after interpolation, that is, the high-resolution image.
  • the row coordinates and column coordinates of the position, H L and W L respectively represent the height and width of the original image
  • H H and W H respectively represent the height and width of the image after interpolation
  • the determining an edge direction of the pixel to be interpolated includes:
  • S1021 Determine a covariance matrix of all pixel gradient weights in the neighborhood of the pixel to be interpolated in the original image window, including:
  • FIG. 2 is a schematic diagram of the Sobel gradient method in the first embodiment of the image scaling method based on the edge adaptive method, as shown in FIG. 2, according to the pixel to be interpolated.
  • H*W the window of H*W in its neighborhood to calculate the edge direction consistency, edge reliability and edge intensity.
  • the gradient calculation uses the Sobel gradient operator method.
  • the horizontal gradient g H and the vertical gradient g V are:
  • S1022 Calculate an eigenvalue and a feature vector of the covariance matrix, and use a feature vector corresponding to the smaller feature value as an edge direction of the pixel to be interpolated;
  • the eigenvector of ⁇ 1 is proportional to the edge direction of the pixel to be interpolated, and the edge direction of the pixel to be interpolated can be expressed as:
  • FIG. 3 is a schematic diagram showing the relationship between the edge direction and the set parameter in the first embodiment of the image scaling method based on edge adaptation according to the present invention, and the slope of the edge direction and The relationship is shown in Figure 3.
  • the edge direction and the set parameter in the first embodiment of the image scaling method based on edge adaptation according to the present invention is shown in Figure 3.
  • the determining the strong edge probability of the pixel to be interpolated comprises:
  • the determining the edge direction reliability parameter according to the covariance matrix of all pixel gradients in the original image window comprises:
  • FIG. 4 is an incremental structure constructed in the first embodiment of the image scaling method based on edge adaptation according to the present invention.
  • FIG. 5 is an edge adaptive image based on the present invention.
  • the relationship between the edge reliability and the constructor is shown in Fig. 5.
  • the relationship between the edge reliability and the constructed function is shown in Fig. 5.
  • determining the edge strength parameter according to the weighted average of all pixel gradient magnitudes in the neighborhood of the pixel to be interpolated in the original image window comprises:
  • w(i,j) represents the bilinear interpolation weight of the (i,j) position pixel
  • g H (i,j) represents the horizontal gradient of the pixel at the (i,j) position
  • g V (i,j) a vertical gradient representing pixels of the (i, j) position
  • T3, and T4 representing a first threshold of edge strength and a second threshold of edge strength, respectively;
  • S10233 Determine an edge direction consistency parameter according to a variance of a direction slope of all pixels in a neighborhood of the pixel to be interpolated in the original image window.
  • determining the edge direction consistency parameter according to a variance of a direction slope of all pixels in a neighborhood of the pixel to be interpolated in the original image window includes:
  • FIG. 6 is a functional relationship diagram of edge direction uniformity and slope variance in the first embodiment of the edge-adaptive image scaling method according to the present invention.
  • the relationship between the edge direction consistency parameter Rc and the calculated slope variance var is as shown in FIG. curve.
  • S10231, S10232, and S10233 are parallel steps, and the order is not distinguished.
  • edge interpolation method refer to the patent application “Image Interpolation System and Method Based on Edge Detection”, for example. In terms of, it can include:
  • the pixel to be interpolated has no direction, and the non-edge image interpolation method is used for interpolation, and the two-dimensional image interpolation is decomposed into two horizontal and vertical images.
  • the one-dimensional direction is sequentially interpolated, and the order of horizontal interpolation and vertical interpolation can be exchanged.
  • the pixel to be interpolated has a direction and is interpolated using the line intersection method.
  • FIG. 7 is a schematic diagram of a line intersection method according to Embodiment 1 of an image adaptive method based on edge adaptation according to the present invention.
  • an intersection of an edge direction and a plurality of rows in a neighborhood of a pixel to be interpolated is calculated.
  • the edge direction is horizontal, there is no intersection with several rows in the neighborhood of the pixel to be interpolated; when the absolute value of the edge direction slope is smaller than the preset threshold k T , it is adjacent to several rows in the neighborhood of the pixel to be interpolated The intersection is farther and the correlation with the pixels to be interpolated is relatively small.
  • the non-edge image interpolation method is used for interpolation, and the two-dimensional image interpolation is decomposed into two horizontal directions and one vertical direction for interpolation, and the order of horizontal interpolation and vertical interpolation can be exchanged.
  • intersection points P 0 , P 1 , P 2 , P of the upper and lower rows of the pixel to be interpolated are taken.
  • the coordinates are:
  • the pixel values of the four intersections are calculated by one-dimensional interpolation in the horizontal direction.
  • the one-dimensional interpolation here can be calculated by any existing image interpolation method. For example, using bilinear interpolation, P 0 in FIG. 9 is taken as an example.
  • the interpolation result of P 0 can be obtained as follows:
  • the pixel values of the four intersection points P 0 , P 1 , P 2 , and P 3 are one-dimensionally filtered to obtain the values of the pixels to be interpolated:
  • I H (i H ,j H ) f 0 *I P0 +f 1 *I P1 +f 2 *I P2 +f 3 *I P3 (22)
  • [f 0 , f 1 , f 2 , f 3 ] are the coefficients of the one-dimensional filter, such as [ 1 , 3 , 3 , 1 ].
  • the non-edge interpolation method can be interpolated by any existing image interpolation method.
  • the two-dimensional image interpolation is decomposed into horizontal and vertical two-dimensional directions for interpolation, and the horizontal interpolation and vertical interpolation can be exchanged.
  • An example of interpolation in the one-dimensional direction is bilinear interpolation, such as the interpolation of P0 in FIG.
  • the fusion weight and the non-edge interpolation fusion weight are used to perform weighted fusion on the interpolation result obtained by the edge interpolation method and the non-edge interpolation method to obtain a final interpolated image, including:
  • the edge adaptive image scaling method of the present invention can use a large number of original points to perform interpolation processing with an arbitrary integer or non-integer integer, and perform interpolation processing in an arbitrary edge direction, so that the enlarged image edges are clear and the sawtooth phenomenon is avoided.
  • FIG. 8 is a structural block diagram of Embodiment 1 of an edge-adaptive image scaling system according to the present invention. As shown in FIG. 8, the edge-adaptive image scaling system of the present invention includes:
  • a coordinate calculation module configured to determine a position of the pixel to be interpolated in the original image
  • a strong edge detection module for calculating an edge direction and a strong edge probability of the pixel to be interpolated
  • An edge interpolation module configured to perform interpolation by using an edge interpolation method according to the edge direction when a strong edge probability of the pixel to be interpolated is not zero;
  • a non-edge interpolation module configured to perform interpolation according to a non-edge interpolation method when a strong edge probability of the pixel to be interpolated is zero;
  • a fusion module configured to use the strong edge probability and the complementary probability of the strong edge probability as edge interpolation fusion weights and non-edge interpolation fusion weights respectively, and the interpolation results obtained by the edge interpolation method and the non-edge interpolation method A weighted fusion is performed to obtain the final interpolated image.
  • the strong edge detection module comprises a gradient calculation module, an edge direction reliability submodule, an edge intensity submodule, and an edge direction consistency submodule:
  • the gradient calculation module is configured to calculate an edge direction of the pixel to be interpolated
  • the edge direction reliability sub-module is configured to determine an edge direction reliability parameter according to a covariance matrix weighted by all pixel gradients in a neighborhood of the pixel to be interpolated in the original image window;
  • the edge strength sub-module is configured to determine an edge strength parameter according to a weighted average of all pixel gradient magnitudes in a neighborhood of the pixel to be interpolated in the original image window;
  • the edge direction consistency sub-module is configured to determine an edge direction consistency parameter according to a variance of a direction slope of all pixels in a neighborhood of the pixel to be interpolated in the original image window;
  • the gradient calculation module is specifically configured to determine a covariance matrix of all pixel gradient weights in the neighborhood of the pixel to be interpolated in the original image window, and calculate eigenvalues and eigenvectors of the covariance matrix, and the smaller features a feature vector corresponding to the value and serving as an edge direction of the pixel to be interpolated;
  • the edge direction reliability sub-module is specifically configured to construct an increasing function f(x) of a ratio of a larger eigenvalue and a smaller eigenvalue of the covariance matrix:
  • ⁇ 1 and ⁇ 2 respectively represent larger eigenvalues and smaller eigenvalues of the covariance matrix
  • A, B, and C represent elements (M) 11 and (M) 12 of the covariance matrix M, respectively.
  • (M) 21 and (M) 22 , T1, T2 represent a first threshold of edge reliability and a second threshold of edge reliability, respectively.
  • the edge intensity sub-module is specifically configured to determine a weighted average value AvgG of all pixel gradient magnitudes in the neighborhood of the pixel to be interpolated in the original image window:
  • w(i,j) represents the bilinear interpolation weight of the (i,j) position pixel
  • g H (i,j) represents the horizontal gradient of the pixel at the (i,j) position
  • g V (i,j) A vertical gradient representing the pixel at the (i, j) position
  • T3, and T4 representing the edge intensity first threshold and the edge intensity second threshold, respectively.
  • the edge direction consistency sub-module is specifically configured to determine a slope variance var of a pixel in a neighborhood of the pixel to be interpolated in the interpolated image:
  • T5 represent the first threshold value of the edge direction consistency and the second threshold of the edge direction consistency, respectively.

Abstract

Provided in the present invention are a method and system for image scaling based on edge self-adaptation. The method comprises: determining the positions of pixels to be interpolated in an original image; calculating an edge direction and a strong-edge probability of the pixels to be interpolated; if the strong-edge probability of the pixels to be interpolated is not equal to zero, then interpolating by utilizing an edge interpolation method on the basis of the edge direction; if the strong-edge probability of the pixels to be interpolated is zero, then interpolating on the basis of a non-edge interpolation method; making the strong-edge probability and a complementary probability of the strong-edge probability respectively as an edge interpolation fusing weight and a non-edge interpolation fusing weight, and performing a weighted fusion on an interpolation result produced by the edge interpolation method and that by the non-edge interpolation method to produce a final interpolated image. The method of the present invention allows image interpolation in any arbitrary edge direction and production of a clear and unpixelated image when scaled.

Description

基于边缘自适应的图像缩放方法及系统Image scaling method and system based on edge adaptation 技术领域Technical field
本发明属于图像处理领域，尤其涉及基于边缘自适应的图像缩放方法及系统。The invention belongs to the field of image processing, and in particular relates to an image scaling method and system based on edge adaptation.
背景技术Background technique
目前，视频技术的主流发展方向之一为超高清技术。将原有的视频源，如标清(720*576)，高清(1920*1080)，通过视频缩放转换成超高清视频，对视频缩放算法的性能要求非常高。At present, one of the mainstream development directions of video technology is ultra high definition technology. The original video source, such as standard definition (720*576), high definition (1920*1080), is converted to ultra high definition video by video scaling, and the performance requirements of the video scaling algorithm are very high.
传统的基于插值的视频缩放方法主要有两类：基于低通滤波的插值方法和基于边缘的插值方法。基于低通滤波的插值方法，如双线性插值，双立方插值，多相位插值等，在获得较平滑的插值图像的同时，会造成图像中高频信息丢失，在图像的边缘出现模糊及锯齿现象。基于边缘的图像插值方法，通过边缘检测，计算出插值点的边缘方向，沿边缘方向对插值点进行插值，可以获得平滑的图像边缘，避免锯齿现象，但是在纹理区域会造成模糊。There are two main types of traditional interpolation-based video scaling methods: low-pass filtering based interpolation and edge-based interpolation. Interpolation methods based on low-pass filtering, such as bilinear interpolation, bicubic interpolation, multi-phase interpolation, etc., while obtaining a smoother interpolated image, will cause high-frequency information loss in the image, blurring and aliasing at the edge of the image. . Based on the edge-based image interpolation method, the edge direction of the interpolation point is calculated by edge detection, and the interpolation points are interpolated along the edge direction to obtain a smooth image edge to avoid aliasing, but it will cause blurring in the texture area.
因此，需要提出一种视频缩放方法，既能保持插值图像中纹理区域的清晰，又能避免插值图像边缘产生的模糊及锯齿现象。Therefore, there is a need to propose a video scaling method that not only keeps the texture region in the interpolated image clear, but also avoids blurring and aliasing caused by the edge of the interpolated image.
发明内容Summary of the invention
本发明提供一种基于边缘自适应的图像缩放方法及系统，以实现提高插值图像的清晰度和减少边缘锯齿的目的。The invention provides an image scaling method and system based on edge adaptation, so as to achieve the purpose of improving the definition of the interpolated image and reducing edge aliasing.
本发明的第一个方面是提供一种基于边缘自适应的图像缩放方法，包括：A first aspect of the present invention provides an image scaling method based on edge adaptation, including:
确定待插值像素在原图像中的位置；Determining the position of the pixel to be interpolated in the original image;
计算待插值像素的边缘方向和强边概率；Calculating the edge direction and the strong edge probability of the pixel to be interpolated;
若所述待插值像素的强边概率不为零，则根据所述边缘方向利用边缘插值法进行插值； If the strong edge probability of the pixel to be interpolated is not zero, interpolation is performed according to the edge direction by using an edge interpolation method;
若所述待插值像素的强边概率为零，则根据非边缘插值法进行插值；If the strong edge probability of the pixel to be interpolated is zero, the interpolation is performed according to the non-edge interpolation method;
将所述强边概率和所述强边概率的互补概率分别作为边缘插值融合权重和非边缘插值融合权重，对所述边缘插值法和所述非边缘插值法获得的插值结果进行加权融合得到最终的插值后图像。Taking the strong edge probability and the complementary probability of the strong edge probability as edge interpolation fusion weights and non-edge interpolation fusion weights respectively, weighting the interpolation results obtained by the edge interpolation method and the non-edge interpolation method to obtain a final result The interpolated image.
本发明的第二个方面是提供一种基于边缘自适应的图像缩放系统，包括：A second aspect of the present invention provides an edge adaptive image scaling system, including:
坐标计算模块，用于确定待插值像素在原图像中的位置；a coordinate calculation module, configured to determine a position of the pixel to be interpolated in the original image;
强边检测模块，用于计算待插值像素的边缘方向和强边概率；a strong edge detection module for calculating an edge direction and a strong edge probability of the pixel to be interpolated;
边缘插值模块，用于在所述待插值像素的强边概率不为零时根据所述边缘方向利用边缘插值法进行插值；An edge interpolation module, configured to perform interpolation by using an edge interpolation method according to the edge direction when a strong edge probability of the pixel to be interpolated is not zero;
非边缘插值模块，用于在所述待插值像素的强边概率为零时根据非边缘插值法进行插值；a non-edge interpolation module, configured to perform interpolation according to a non-edge interpolation method when a strong edge probability of the pixel to be interpolated is zero;
融合模块，用于将所述强边概率和所述强边概率的互补概率分别作为边缘插值融合权重和非边缘插值融合权重，对所述边缘插值法和所述非边缘插值法获得的插值结果进行加权融合得到最终的插值后图像。a fusion module, configured to use the strong edge probability and the complementary probability of the strong edge probability as edge interpolation fusion weights and non-edge interpolation fusion weights respectively, and the interpolation results obtained by the edge interpolation method and the non-edge interpolation method A weighted fusion is performed to obtain the final interpolated image.
本发明的有益效果为：The beneficial effects of the invention are:
本发明基于边缘自适应的图像缩放方法能够使用数量较大的原始点以任意整数或非整数缩放倍率、及在任意边缘方向进行插值处理，使放大后的图像边缘清晰且避免了锯齿现象。The edge adaptive image scaling method of the present invention can use a large number of original points to perform interpolation processing with an arbitrary integer or non-integer integer, and perform interpolation processing in an arbitrary edge direction, so that the enlarged image edges are clear and the sawtooth phenomenon is avoided.
附图说明DRAWINGS
图1为本发明基于边缘自适应的图像缩放方法实施例一的流程图；1 is a flowchart of Embodiment 1 of an image adaptive method based on edge adaptation according to the present invention;
图2为本发明基于边缘自适应的图像缩放方法实施例一中Sobel梯度方法的示意图；2 is a schematic diagram of a Sobel gradient method according to Embodiment 1 of an edge-adaptive image scaling method according to the present invention;
图3为本发明基于边缘自适应的图像缩放方法实施例一中边缘方向与设定参数间关系的示意图；3 is a schematic diagram showing the relationship between edge directions and setting parameters in the first embodiment of the image scaling method based on edge adaptation according to the present invention;
图4为本发明基于边缘自适应的图像缩放方法实施例一中构造的递增函数关系图；4 is an incremental function relationship diagram constructed in Embodiment 1 of an edge adaptive image scaling method according to the present invention;
图5为本发明基于边缘自适应的图像缩放方法实施例一中边缘可靠度与构造函数的函数关系图；FIG. 5 is a functional relationship diagram between edge reliability and a constructor in the first embodiment of the image scaling method based on edge adaptation according to the present invention; FIG.
图6为本发明基于边缘自适应的图像缩放方法实施例一中边缘方向 一致性与斜率方差的函数关系图；FIG. 6 is an edge direction of the first embodiment of the image scaling method based on edge adaptation according to the present invention; a functional relationship diagram of consistency and slope variance;
图7为本发明基于边缘自适应的图像缩放方法实施例一中行交点法的示意图；7 is a schematic diagram of a line intersection method according to Embodiment 1 of an edge adaptive image scaling method according to the present invention;
图8为本发明基于边缘自适应的图像缩放系统实施例一的结构框图。FIG. 8 is a structural block diagram of Embodiment 1 of an image scaling system based on edge adaptation according to the present invention.
具体实施方式detailed description
图1为本发明基于边缘自适应的图像缩放方法实施例一的流程图，如图1所示，本发明基于边缘自适应的图像缩放方法包括：FIG. 1 is a flowchart of Embodiment 1 of an edge-adaptive image scaling method according to the present invention. As shown in FIG. 1 , the edge adaptive image scaling method of the present invention includes:
S101、根据原图像和插值后图像的大小，确定待插值像素在原图像中的位置，包括：S101. Determine, according to the original image and the size of the interpolated image, a position of the pixel to be interpolated in the original image, including:
假设输入为原图像也即低分辨率图像IL，其图像宽和高分别为WL和HL，插值后的高分辨率图像为IH，其图像宽和高分别为WH和HH，对于插值图像也即高分辨率图像中位于iH行jH列坐标为(iH,jH)的待插值像素，通过坐标计算单元，可以得到其在低分辨率图像中的坐标(iL,jL)：Assume that the input is the original image, that is, the low-resolution image I L , the image width and height are W L and H L , respectively, and the interpolated high-resolution image is I H , and the image width and height are W H and H H , respectively. For the interpolated image, that is, the pixel to be interpolated in the i H line j H column coordinate (i H , j H ) in the high-resolution image, the coordinate in the low-resolution image can be obtained by the coordinate calculation unit (i L , j L ):
Figure PCTCN2015075696-appb-000001
Figure PCTCN2015075696-appb-000001
其中，iL和jL分别表示待插值像素在原图像也即低分辨率图像中位置的行坐标和列坐标，iH和jH分别表示待插值像素在插值后图像也即高分辨率图像中位置的行坐标和列坐标，HL和WL分别表示原图像的高和宽，HH和WH分别表示插值后图像的高和宽；Wherein, i L and j L respectively represent row coordinates and column coordinates of the position of the pixel to be interpolated in the original image, that is, the low-resolution image, and i H and j H respectively represent the pixel to be interpolated in the image after interpolation, that is, the high-resolution image. The row coordinates and column coordinates of the position, H L and W L respectively represent the height and width of the original image, and H H and W H respectively represent the height and width of the image after interpolation;
S102、计算待插值像素的边缘方向和强边概率；S102. Calculate an edge direction and a strong edge probability of the pixel to be interpolated.
优选的，所述确定待插值像素的边缘方向包括：Preferably, the determining an edge direction of the pixel to be interpolated includes:
S1021、确定原图像窗口中待插值像素邻域内全部像素梯度加权的协方差矩阵，包括：S1021: Determine a covariance matrix of all pixel gradient weights in the neighborhood of the pixel to be interpolated in the original image window, including:
首先计算原图像上像素点的梯度，然后计算边缘方向v；图2为本发明基于边缘自适应的图像缩放方法实施例一中Sobel梯度方法的示意图，如图2所示，根据待插值像素在原图像中的位置，选取其邻域内H*W的窗口，进行边缘方向一致性、边缘可靠度和边缘强度的计算，本例中H＝4，W＝6；最终的强边概率为边缘方向一致性、边缘可靠度和边缘强度三者的乘积；First, the gradient of the pixel points on the original image is calculated, and then the edge direction v is calculated. FIG. 2 is a schematic diagram of the Sobel gradient method in the first embodiment of the image scaling method based on the edge adaptive method, as shown in FIG. 2, according to the pixel to be interpolated. For the position in the image, select the window of H*W in its neighborhood to calculate the edge direction consistency, edge reliability and edge intensity. In this example, H=4, W=6; the final strong edge probability is the same in the edge direction. Product of sex, edge reliability and edge strength;
梯度计算采用Sobel梯度算子的方法，对于原图像上的像素(i,j)，其水 平梯度gH和垂直梯度gV分别为：The gradient calculation uses the Sobel gradient operator method. For the pixel (i, j) on the original image, the horizontal gradient g H and the vertical gradient g V are:
gH(i,j)＝IL(i-1,j+1)+2*IL(i,j+1)+IL(i+1,j+1)g H (i,j)=I L (i-1,j+1)+2*I L (i,j+1)+I L (i+1,j+1)
-IL(i-1,j-1)-2*IL(i,j-1)-IL(i+1,j-1)     (2)-I L (i-1,j-1)-2*I L (i,j-1)-I L (i+1,j-1) (2)
gV(i,j)＝IL(i-1,j-1)+2*IL(i-1,j)+IL(i-1,j+1)g V (i,j)=I L (i-1,j-1)+2*I L (i-1,j)+I L (i-1,j+1)
-IL(i+1,j-1)-2*IL(i+1,j)-IL(i+1,j+1)    (3)-I L (i+1,j-1)-2*I L (i+1,j)-I L (i+1,j+1) (3)
再根据待插值像素的位置，计算其邻域内所有像素梯度加权的协方差矩阵：Then, according to the position of the pixel to be interpolated, the covariance matrix of all pixel gradient weights in the neighborhood is calculated:
Figure PCTCN2015075696-appb-000002
Figure PCTCN2015075696-appb-000002
其中，w(i,j)的取值如表1所示：Where w(i,j) has the values shown in Table 1:
(1-dx)*(1-d(1-dx)*(1-d (1-dy)(1-dy) (1-dy)(1-dy) (1-dy)(1-dy) (1-dy)(1-dy) dx*(1-dDx*(1-d
(1-dx)(1-dx) 11 11 11 11 dxDx
(1-dx)(1-dx) 11 11 11 11 dxDx
(1-dx)*dy(1-dx)*dy dyDy dyDy dyDy dyDy dx*dyDx*dy
表1Table 1
S1022、计算所述协方差矩阵的特征值和特征向量，并将较小特征值对应的特征向量作为待插值像素的边缘方向；S1022: Calculate an eigenvalue and a feature vector of the covariance matrix, and use a feature vector corresponding to the smaller feature value as an edge direction of the pixel to be interpolated;
假设协方差矩阵M的两个特征值为λ1和λ2，其中λ2为较大的特征值，Suppose that the two eigenvalues of the covariance matrix M are λ1 and λ2, where λ2 is a large eigenvalue,
Figure PCTCN2015075696-appb-000003
Figure PCTCN2015075696-appb-000003
Figure PCTCN2015075696-appb-000004
Figure PCTCN2015075696-appb-000004
λ1的特征向量与待插值像素的边缘方向成比例,待插值像素的边缘方向可以表示为：The eigenvector of λ1 is proportional to the edge direction of the pixel to be interpolated, and the edge direction of the pixel to be interpolated can be expressed as:
Figure PCTCN2015075696-appb-000005
Figure PCTCN2015075696-appb-000005
由于特征值的计算中有开根号操作，不利于硬件实现，下面提供一种求取边缘方向优化方法；Since the root value operation is performed in the calculation of the feature value, which is not conducive to hardware implementation, the following provides an edge direction optimization method;
若B为0，可以推出边缘方向为以下三种情况：If B is 0, the following three cases can be introduced:
B＝0，A＝C，无方向；B=0, A=C, no direction;
B＝0，A>C，垂直方向； B=0, A>C, vertical direction;
B＝0，A<C，水平方向B=0, A<C, horizontal direction
若B不为0，边缘方向的斜率可以表示为：If B is not 0, the slope of the edge direction can be expressed as:
Figure PCTCN2015075696-appb-000006
Figure PCTCN2015075696-appb-000006
Figure PCTCN2015075696-appb-000007
Figure PCTCN2015075696-appb-000007
图3为本发明基于边缘自适应的图像缩放方法实施例一中边缘方向与设定参数间关系的示意图，边缘方向的斜率与
Figure PCTCN2015075696-appb-000008
的关系如图3所示，从图中可以看出，四条曲线之间存在着一定关系，或对称，或乘积为-1，因此，只需要将其中的一条曲线量化，做成查找表即可，计算
Figure PCTCN2015075696-appb-000009
通过查表得到对应的斜率值，得到待插值像素的边缘方向；FIG. 3 is a schematic diagram showing the relationship between the edge direction and the set parameter in the first embodiment of the image scaling method based on edge adaptation according to the present invention, and the slope of the edge direction and
Figure PCTCN2015075696-appb-000008
The relationship is shown in Figure 3. As can be seen from the figure, there is a certain relationship between the four curves, or symmetry, or the product is -1. Therefore, only one of the curves needs to be quantized to make a lookup table. , calculation
Figure PCTCN2015075696-appb-000009
Obtaining a corresponding slope value by looking up the table to obtain an edge direction of the pixel to be interpolated;
优选的，所述确定待插值像素的强边概率包括：Preferably, the determining the strong edge probability of the pixel to be interpolated comprises:
S1023、确定下述三个参数：S1023, determining the following three parameters:
S10231、根据原图像窗口内全部像素梯度加权的协方差矩阵确定边缘方向可靠度参数；S10231. Determine an edge direction reliability parameter according to a covariance matrix weighted by all pixel gradients in the original image window.
优选的，所述根据原图像窗口内全部像素梯度的协方差矩阵确定边缘方向可靠度参数包括：Preferably, the determining the edge direction reliability parameter according to the covariance matrix of all pixel gradients in the original image window comprises:
构造所述协方差矩阵的较大特征值和较小特征值的比值的递增函数f(x):An incremental function f(x) that constructs a ratio of a larger eigenvalue of the covariance matrix to a smaller eigenvalue:
Figure PCTCN2015075696-appb-000010
Figure PCTCN2015075696-appb-000010
根据公式(11)确定边缘方向可靠度参数:Determine the edge direction reliability parameter according to formula (11):
Figure PCTCN2015075696-appb-000011
Figure PCTCN2015075696-appb-000011
其中，λ1、λ2分别表示所述协方差矩阵的较大特征值和较小特征值，A、B、和C分别表示所述协方差矩阵M的元素(M)11、(M)12或(M)21、和(M)22，T1、T2分别表示预先设定的边缘可靠度第一阈值和边缘可靠度第二阈值； 需要说明的是，两特征值的比值
Figure PCTCN2015075696-appb-000012
可表征边缘的可靠度，为了避免特征值计算时的开根号操作，构造一个关于两特征值比值的递增函数，图4为本发明基于边缘自适应的图像缩放方法实施例一中构造的递增函数关系图，由图4所示曲线可以看出，构造的函数关于两特征值比值单调递增，因此可以通过计算构造的函数进行边缘可靠度的计算，图5为本发明基于边缘自适应的图像缩放方法实施例一中边缘可靠度与构造函数的函数关系图，边缘可靠度与构造的函数的关系如图5所示。Where λ 1 and λ 2 respectively represent larger eigenvalues and smaller eigenvalues of the covariance matrix, and A, B, and C represent elements (M) 11 and (M) 12 of the covariance matrix M, respectively. Or (M) 21 , and (M) 22 , T1, T2 respectively represent a preset first threshold of edge reliability and a second threshold of edge reliability; respectively, the ratio of the two eigenvalues is
Figure PCTCN2015075696-appb-000012
The reliability of the edge can be characterized. In order to avoid the root opening operation in the calculation of the eigenvalue, an increasing function about the ratio of the two eigenvalues is constructed. FIG. 4 is an incremental structure constructed in the first embodiment of the image scaling method based on edge adaptation according to the present invention. The function relationship diagram, as can be seen from the curve shown in Fig. 4, the constructed function monotonically increases with respect to the ratio of the two eigenvalues, so the edge reliability can be calculated by calculating the constructed function. FIG. 5 is an edge adaptive image based on the present invention. In the first embodiment of the scaling method, the relationship between the edge reliability and the constructor is shown in Fig. 5. The relationship between the edge reliability and the constructed function is shown in Fig. 5.
S10232、根据原图像窗口中待插值像素邻域内全部像素梯度幅值的加权平均确定边缘强度参数；S10232. Determine an edge strength parameter according to a weighted average of all pixel gradient magnitudes in a neighborhood of the pixel to be interpolated in the original image window.
优选的，所述根据原图像窗口中待插值像素邻域内全部像素梯度幅值的加权平均确定边缘强度参数包括：Preferably, determining the edge strength parameter according to the weighted average of all pixel gradient magnitudes in the neighborhood of the pixel to be interpolated in the original image window comprises:
确定原图像窗口中待插值像素邻域内全部像素梯度幅值的加权平均值AvgG：Determining the weighted average AvgG of all pixel gradient magnitudes in the neighborhood of the pixel to be interpolated in the original image window:
Figure PCTCN2015075696-appb-000013
Figure PCTCN2015075696-appb-000013
根据公式(13)确定边缘强度Rs：Determine the edge strength Rs according to formula (13):
Figure PCTCN2015075696-appb-000014
Figure PCTCN2015075696-appb-000014
其中，w(i,j)表示(i,j)位置像素的双线性插值权重，gH(i,j)表示(i,j)位置的像素的水平梯度，gV(i,j)表示(i,j)位置的像素的垂直梯度，T3、和T4分别表示边缘强度第一阈值和边缘强度第二阈值；Where w(i,j) represents the bilinear interpolation weight of the (i,j) position pixel, g H (i,j) represents the horizontal gradient of the pixel at the (i,j) position, g V (i,j) a vertical gradient representing pixels of the (i, j) position, T3, and T4 representing a first threshold of edge strength and a second threshold of edge strength, respectively;
S10233、根据原图像窗口中待插值像素邻域内全部像素的方向斜率的方差确定边缘方向一致性参数；S10233. Determine an edge direction consistency parameter according to a variance of a direction slope of all pixels in a neighborhood of the pixel to be interpolated in the original image window.
优选的，所述根据原图像窗口中待插值像素邻域内全部像素的方向斜率的方差确定边缘方向一致性参数包括：Preferably, determining the edge direction consistency parameter according to a variance of a direction slope of all pixels in a neighborhood of the pixel to be interpolated in the original image window includes:
确定插值图像中待插值像素邻域内像素的的斜率方差var：Determining the slope variance var of the pixels in the neighborhood of the pixel to be interpolated in the interpolated image:
Figure PCTCN2015075696-appb-000015
Figure PCTCN2015075696-appb-000015
根据公式(15)确定边缘方向一致性参数：Determine the edge direction consistency parameter according to formula (15):
Figure PCTCN2015075696-appb-000016
Figure PCTCN2015075696-appb-000016
其中，
Figure PCTCN2015075696-appb-000017
表示待插值像素邻域内像素的 方向斜率，
Figure PCTCN2015075696-appb-000018
表示待插值像素领域内像素的方向斜率的平均值，T5、和T6分别表示边缘方向一致性第一阈值和边缘方向一致性第二阈值；需要说明的是，在确定AvgG时，需在原图像窗口上，取当前待插值像素的P*Q邻域S，计算邻域内有方向像素的方向斜率方差，有方向是指该像素的方向矢量v的水平分量vx和垂直分量vy不同时为0，邻域内有方向像素的个数记为N，由于垂直方向的斜率为无穷大，这里对斜率的最大值进行限制，如果斜率的绝对值大于预先设定的阈值T，则令其绝对值等于T；图6为本发明基于边缘自适应的图像缩放方法实施例一中边缘方向一致性与斜率方差的函数关系图，边缘方向一致性参数Rc与计算出的斜率方差var的关系为图6所示曲线。among them,
Figure PCTCN2015075696-appb-000017
 Indicates the slope of the direction of the pixel in the neighborhood of the pixel to be interpolated,
Figure PCTCN2015075696-appb-000018
The average value of the direction slopes of the pixels in the field of the pixel to be interpolated is represented, and T5, and T6 respectively represent the first threshold value of the edge direction consistency and the second threshold value of the edge direction consistency; it should be noted that in determining the AvgG, the original image window is required. Upper, take the P*Q neighborhood S of the current pixel to be interpolated, and calculate the direction slope variance of the directional pixel in the neighborhood. The direction is that the horizontal component v x and the vertical component v y of the direction vector v of the pixel are different at 0. The number of directional pixels in the neighborhood is recorded as N. Since the slope in the vertical direction is infinite, the maximum value of the slope is limited here. If the absolute value of the slope is greater than the preset threshold T, the absolute value is equal to T. FIG. 6 is a functional relationship diagram of edge direction uniformity and slope variance in the first embodiment of the edge-adaptive image scaling method according to the present invention. The relationship between the edge direction consistency parameter Rc and the calculated slope variance var is as shown in FIG. curve.
需要说明的是，S10231、S10232、S10233为并行步骤，不区分先后顺序。It should be noted that S10231, S10232, and S10233 are parallel steps, and the order is not distinguished.
S1024、将所确定的所述三个参数分别作为所述强边融合参数的相互独立的乘性因子相乘得到待插值像素的强边概率，也即得到最终强边概率Rel：S1024. Multiplying the determined three parameters as independent multiplicative multiplicative factors of the strong edge fusion parameter respectively to obtain a strong edge probability of the pixel to be interpolated, that is, obtaining a final strong edge probability Rel:
Rel＝Ra*Rs*Rc           (16)Rel=Ra*Rs*Rc (16)
S103、若所述待插值像素的强边概率不为零，则根据所述边缘方向利用边缘插值法进行插值，所述边缘插值法参见专利申请“基于边缘检测的图像插值系统及方法”，举例来说，可以包括：S103. If the strong edge probability of the pixel to be interpolated is not zero, perform interpolation according to the edge direction by using an edge interpolation method. For the edge interpolation method, refer to the patent application “Image Interpolation System and Method Based on Edge Detection”, for example. In terms of, it can include:
若待插值像素的边缘方向的水平分量vx和垂直分量vy都为0，则待插值像素无方向，采用非边缘的图像插值方法进行插值，将二维图像插值分解为水平和垂直两个一维方向依次进行插值，水平方向插值和垂直方向插值的顺序可交换。If the horizontal component v x and the vertical component v y of the edge direction of the pixel to be interpolated are both 0, the pixel to be interpolated has no direction, and the non-edge image interpolation method is used for interpolation, and the two-dimensional image interpolation is decomposed into two horizontal and vertical images. The one-dimensional direction is sequentially interpolated, and the order of horizontal interpolation and vertical interpolation can be exchanged.
若待插值像素的边缘方向的水平分量vx和垂直分量vy不都为0，则待插值像素有方向，使用行交点方法插值。If the horizontal component v x and the vertical component v y of the edge direction of the pixel to be interpolated are not all 0, the pixel to be interpolated has a direction and is interpolated using the line intersection method.
图7为本发明基于边缘自适应的图像缩放方法实施例一中行交点法的示意图，如图7所示，计算边缘方向与待插值像素邻域内若干行的交点。当边缘方向为水平方向时，其与待插值像素邻域内若干行没有交点；当边缘方向斜率的绝对值较小，小于预先设定的阈值kT时，其与待插值像素邻域内若干行的交点较远，与待插值像素的相关性相对较小。因此，对上述 两种情况采用非边缘的图像插值方法进行插值，将二维图像插值分解为水平和垂直两个一维方向依次进行插值，水平方向插值和垂直方向插值的顺序可交换。FIG. 7 is a schematic diagram of a line intersection method according to Embodiment 1 of an image adaptive method based on edge adaptation according to the present invention. As shown in FIG. 7, an intersection of an edge direction and a plurality of rows in a neighborhood of a pixel to be interpolated is calculated. When the edge direction is horizontal, there is no intersection with several rows in the neighborhood of the pixel to be interpolated; when the absolute value of the edge direction slope is smaller than the preset threshold k T , it is adjacent to several rows in the neighborhood of the pixel to be interpolated The intersection is farther and the correlation with the pixels to be interpolated is relatively small. Therefore, for the above two cases, the non-edge image interpolation method is used for interpolation, and the two-dimensional image interpolation is decomposed into two horizontal directions and one vertical direction for interpolation, and the order of horizontal interpolation and vertical interpolation can be exchanged.
对于上述两种情况以外的其他情况，找到边缘方向与待插值像素邻域内若干行的交点，本实施例中取待插值像素上下两行共四行的交点P0，P1，P2，P3，坐标分别为：For other cases than the above two cases, find the intersection of the edge direction and several rows in the neighborhood of the pixel to be interpolated. In this embodiment, the intersection points P 0 , P 1 , P 2 , P of the upper and lower rows of the pixel to be interpolated are taken. 3 , the coordinates are:
Figure PCTCN2015075696-appb-000019
Figure PCTCN2015075696-appb-000019
Figure PCTCN2015075696-appb-000020
Figure PCTCN2015075696-appb-000020
Figure PCTCN2015075696-appb-000021
Figure PCTCN2015075696-appb-000021
Figure PCTCN2015075696-appb-000022
Figure PCTCN2015075696-appb-000022
四个交点的像素值采用水平方向的一维插值计算，这里的一维插值可以采用现有的任一种图像插值方法计算，如使用双线性插值，以图9中的P0为例，可以得到P0的插值结果为：The pixel values of the four intersections are calculated by one-dimensional interpolation in the horizontal direction. The one-dimensional interpolation here can be calculated by any existing image interpolation method. For example, using bilinear interpolation, P 0 in FIG. 9 is taken as an example. The interpolation result of P 0 can be obtained as follows:
Figure PCTCN2015075696-appb-000023
Figure PCTCN2015075696-appb-000023
再对四个交点P0，P1，P2，P3的像素值进行一维滤波，得到待插值像素的值：Then, the pixel values of the four intersection points P 0 , P 1 , P 2 , and P 3 are one-dimensionally filtered to obtain the values of the pixels to be interpolated:
IH(iH,jH)＝f0*IP0+f1*IP1+f2*IP2+f3*IP3      (22)I H (i H ,j H )=f 0 *I P0 +f 1 *I P1 +f 2 *I P2 +f 3 *I P3 (22)
其中，[f0,f1,f2,f3]为一维滤波器的系数，如[1，3，3，1]。Where [f 0 , f 1 , f 2 , f 3 ] are the coefficients of the one-dimensional filter, such as [ 1 , 3 , 3 , 1 ].
S104、若所述待插值像素的强边概率为零，则根据非边缘插值法进行插值；S104. If the strong edge probability of the pixel to be interpolated is zero, perform interpolation according to a non-edge interpolation method;
非边缘插值法可以采用现有的任一种图像插值方法进行插值，如将二维图像插值分解为水平和垂直两个一维方向依次进行插值，水平方向插值和垂直方向插值的顺序可交换，一维方向上插值的一个例子为双线性插值，如在图7中对P0的插值。The non-edge interpolation method can be interpolated by any existing image interpolation method. For example, the two-dimensional image interpolation is decomposed into horizontal and vertical two-dimensional directions for interpolation, and the horizontal interpolation and vertical interpolation can be exchanged. An example of interpolation in the one-dimensional direction is bilinear interpolation, such as the interpolation of P0 in FIG.
S105、将所述强边概率和所述强边概率的互补概率分别作为边缘插值 融合权重和非边缘插值融合权重，对所述边缘插值法和所述非边缘插值法获得的插值结果进行加权融合得到最终的插值后图像，包括：S105. Using the strong edge probability and the complementary probability of the strong edge probability as edge interpolation respectively The fusion weight and the non-edge interpolation fusion weight are used to perform weighted fusion on the interpolation result obtained by the edge interpolation method and the non-edge interpolation method to obtain a final interpolated image, including:
根据公式(1)对所述边缘插值法和所述非边缘插值法获得的插值结果IHe和IHt进行加权融合，得到最终的插值后图像IH：Performing weighted fusion of the interpolation results I He and I Ht obtained by the edge interpolation method and the non-edge interpolation method according to formula (1) to obtain a final interpolated image I H :
IH＝Rel*IHe+(1-Rel)*IHt     (23)I H =Rel*I He +(1-Rel)*I Ht (23)
本发明基于边缘自适应的图像缩放方法能够使用数量较大的原始点以任意整数或非整数缩放倍率、及在任意边缘方向进行插值处理，使放大后的图像边缘清晰且避免了锯齿现象。The edge adaptive image scaling method of the present invention can use a large number of original points to perform interpolation processing with an arbitrary integer or non-integer integer, and perform interpolation processing in an arbitrary edge direction, so that the enlarged image edges are clear and the sawtooth phenomenon is avoided.
图8为本发明基于边缘自适应的图像缩放系统实施例一的结构框图，如图8所示，本发明基于边缘自适应的图像缩放系统包括：FIG. 8 is a structural block diagram of Embodiment 1 of an edge-adaptive image scaling system according to the present invention. As shown in FIG. 8, the edge-adaptive image scaling system of the present invention includes:
坐标计算模块，用于确定待插值像素在原图像中的位置；a coordinate calculation module, configured to determine a position of the pixel to be interpolated in the original image;
强边检测模块，用于计算待插值像素的边缘方向和强边概率；a strong edge detection module for calculating an edge direction and a strong edge probability of the pixel to be interpolated;
边缘插值模块，用于在所述待插值像素的强边概率不为零时根据所述边缘方向利用边缘插值法进行插值；An edge interpolation module, configured to perform interpolation by using an edge interpolation method according to the edge direction when a strong edge probability of the pixel to be interpolated is not zero;
非边缘插值模块，用于在所述待插值像素的强边概率为零时根据非边缘插值法进行插值；a non-edge interpolation module, configured to perform interpolation according to a non-edge interpolation method when a strong edge probability of the pixel to be interpolated is zero;
融合模块，用于将所述强边概率和所述强边概率的互补概率分别作为边缘插值融合权重和非边缘插值融合权重，对所述边缘插值法和所述非边缘插值法获得的插值结果进行加权融合得到最终的插值后图像。a fusion module, configured to use the strong edge probability and the complementary probability of the strong edge probability as edge interpolation fusion weights and non-edge interpolation fusion weights respectively, and the interpolation results obtained by the edge interpolation method and the non-edge interpolation method A weighted fusion is performed to obtain the final interpolated image.
优选的，所述强边检测模块包括梯度计算模块、边缘方向可靠度子模块、边缘强度子模块、和边缘方向一致性子模块：Preferably, the strong edge detection module comprises a gradient calculation module, an edge direction reliability submodule, an edge intensity submodule, and an edge direction consistency submodule:
所述梯度计算模块，用于计算待插值像素的边缘方向；The gradient calculation module is configured to calculate an edge direction of the pixel to be interpolated;
所述边缘方向可靠度子模块，用于根据原图像窗口中待插值像素邻域内全部像素梯度加权的协方差矩阵确定边缘方向可靠度参数；The edge direction reliability sub-module is configured to determine an edge direction reliability parameter according to a covariance matrix weighted by all pixel gradients in a neighborhood of the pixel to be interpolated in the original image window;
所述边缘强度子模块，用于根据原图像窗口中待插值像素邻域内全部像素梯度幅值的加权平均确定边缘强度参数；The edge strength sub-module is configured to determine an edge strength parameter according to a weighted average of all pixel gradient magnitudes in a neighborhood of the pixel to be interpolated in the original image window;
所述边缘方向一致性子模块，用于根据原图像窗口中待插值像素邻域内全部像素的方向斜率的方差确定边缘方向一致性参数；The edge direction consistency sub-module is configured to determine an edge direction consistency parameter according to a variance of a direction slope of all pixels in a neighborhood of the pixel to be interpolated in the original image window;
将所确定的所述参数中的一个或其任意组合分别作为所述强边融合 参数的相互独立的乘性因子相乘得到待插值像素的强边概率。Combining one of the determined parameters or any combination thereof as the strong edge fusion The multiplicative multiplicative factors of the parameters are multiplied to obtain the strong edge probability of the pixel to be interpolated.
优选的，所述梯度计算模块，具体用于确定原图像窗口中待插值像素邻域内全部像素梯度加权的协方差矩阵、和计算所述协方差矩阵的特征值和特征向量，并将较小特征值对应的特征向量并作为待插值像素的边缘方向；Preferably, the gradient calculation module is specifically configured to determine a covariance matrix of all pixel gradient weights in the neighborhood of the pixel to be interpolated in the original image window, and calculate eigenvalues and eigenvectors of the covariance matrix, and the smaller features a feature vector corresponding to the value and serving as an edge direction of the pixel to be interpolated;
所述边缘方向可靠度子模块，具体用于构造所述协方差矩阵的较大特征值和较小特征值的比值的递增函数f(x):The edge direction reliability sub-module is specifically configured to construct an increasing function f(x) of a ratio of a larger eigenvalue and a smaller eigenvalue of the covariance matrix:
Figure PCTCN2015075696-appb-000024
Figure PCTCN2015075696-appb-000024
和根据公式(11)确定边缘方向可靠度参数Ra:And determining the edge direction reliability parameter Ra according to formula (11):
Figure PCTCN2015075696-appb-000025
Figure PCTCN2015075696-appb-000025
其中，λ1、λ2分别表示所述协方差矩阵的较大特征值和较小特征值，A、B、和C分别表示所述协方差矩阵M的元素(M)11、(M)12或(M)21、和(M)22，T1、T2分别表示边缘可靠度第一阈值和边缘可靠度第二阈值。Where λ 1 and λ 2 respectively represent larger eigenvalues and smaller eigenvalues of the covariance matrix, and A, B, and C represent elements (M) 11 and (M) 12 of the covariance matrix M, respectively. Or (M) 21 and (M) 22 , T1, T2 represent a first threshold of edge reliability and a second threshold of edge reliability, respectively.
优选的，所述边缘强度子模块具体用于确定原图像窗口中待插值像素邻域内全部像素梯度幅值的加权平均值AvgG：Preferably, the edge intensity sub-module is specifically configured to determine a weighted average value AvgG of all pixel gradient magnitudes in the neighborhood of the pixel to be interpolated in the original image window:
Figure PCTCN2015075696-appb-000026
Figure PCTCN2015075696-appb-000026
和根据公式(13)确定边缘强度Rs：And determining the edge strength Rs according to formula (13):
Figure PCTCN2015075696-appb-000027
Figure PCTCN2015075696-appb-000027
其中，w(i,j)表示(i,j)位置像素的双线性插值权重，gH(i,j)表示(i,j)位置的像素的水平梯度，gV(i,j)表示(i,j)位置的像素的垂直梯度，T3、和T4分别表示边缘强度第一阈值和边缘强度第二阈值。Where w(i,j) represents the bilinear interpolation weight of the (i,j) position pixel, g H (i,j) represents the horizontal gradient of the pixel at the (i,j) position, g V (i,j) A vertical gradient representing the pixel at the (i, j) position, T3, and T4 representing the edge intensity first threshold and the edge intensity second threshold, respectively.
优选的，所述边缘方向一致性子模块具体用于确定插值图像中待插值像素邻域内像素的斜率方差var：Preferably, the edge direction consistency sub-module is specifically configured to determine a slope variance var of a pixel in a neighborhood of the pixel to be interpolated in the interpolated image:
Figure PCTCN2015075696-appb-000028
Figure PCTCN2015075696-appb-000028
和根据公式(15)确定边缘方向一致性参数Rc：And determining the edge direction consistency parameter Rc according to formula (15):
Figure PCTCN2015075696-appb-000029
Figure PCTCN2015075696-appb-000029
其中，
Figure PCTCN2015075696-appb-000030
表示待插值像素邻域内像素的方向斜率，
Figure PCTCN2015075696-appb-000031
表示待插值像素领域内像素的方向斜率的平均值，T5、和T6分别表示边缘方向一致性第一阈值和边缘方向一致性第二阈值。among them,
Figure PCTCN2015075696-appb-000030
 Representing the slope of the direction of the pixel in the neighborhood of the pixel to be interpolated,
Figure PCTCN2015075696-appb-000031
The average value of the direction slopes of the pixels in the field of the pixel to be interpolated is represented, and T5, and T6 represent the first threshold value of the edge direction consistency and the second threshold of the edge direction consistency, respectively.
最后应说明的是：以上各实施例仅用以说明本发明的技术方案，而非对其限制；尽管参照前述各实施例对本发明进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分或者全部技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本发明各实施例技术方案的范围。 Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, and are not intended to be limiting; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art will understand that The technical solutions described in the foregoing embodiments may be modified, or some or all of the technical features may be equivalently replaced; and the modifications or substitutions do not deviate from the technical solutions of the embodiments of the present invention. range.

Claims (10)

  1. 一种基于边缘自适应的图像缩放方法，其特征在于，包括：An image scaling method based on edge adaptation, comprising:
    确定待插值像素在原图像中的位置；Determining the position of the pixel to be interpolated in the original image;
    计算待插值像素的边缘方向和强边概率；Calculating the edge direction and the strong edge probability of the pixel to be interpolated;
    若所述待插值像素的强边概率不为零，则根据所述边缘方向利用边缘插值法进行插值；If the strong edge probability of the pixel to be interpolated is not zero, interpolation is performed according to the edge direction by using an edge interpolation method;
    若所述待插值像素的强边概率为零，则根据非边缘插值法进行插值；If the strong edge probability of the pixel to be interpolated is zero, the interpolation is performed according to the non-edge interpolation method;
    将所述强边概率和所述强边概率的互补概率分别作为边缘插值融合权重和非边缘插值融合权重，对所述边缘插值法和所述非边缘插值法获得的插值结果进行加权融合得到最终的插值后图像。Taking the strong edge probability and the complementary probability of the strong edge probability as edge interpolation fusion weights and non-edge interpolation fusion weights respectively, weighting the interpolation results obtained by the edge interpolation method and the non-edge interpolation method to obtain a final result The interpolated image.
  2. 根据权利要求1所述的基于边缘自适应的图像缩放方法，其特征在于，所述确定待插值像素的强边概率包括：The edge-adaptive-based image scaling method according to claim 1, wherein the determining a strong edge probability of the pixel to be interpolated comprises:
    确定下述参数中的至少一个参数：Determine at least one of the following parameters:
    根据原图像窗口中待插值像素邻域内全部像素梯度加权的协方差矩阵确定边缘方向可靠度参数；Determining an edge direction reliability parameter according to a covariance matrix weighted by all pixel gradients in a neighborhood of the pixel to be interpolated in the original image window;
    根据原图像窗口中待插值像素邻域内全部像素梯度幅值的加权平均确定边缘强度参数；Determining an edge strength parameter according to a weighted average of all pixel gradient magnitudes in a neighborhood of the pixel to be interpolated in the original image window;
    根据原图像窗口中待插值像素邻域内全部像素的方向斜率的方差确定边缘方向一致性参数；Determining an edge direction consistency parameter according to a variance of a slope of a direction of all pixels in a neighborhood of the pixel to be interpolated in the original image window;
    将所确定的所述参数中的一个或其任意组合分别作为所述强边融合参数的相互独立的乘性因子相乘得到待插值像素的强边概率。Multiplying one of the determined parameters or any combination thereof as mutually independent multiplicative factors of the strong edge fusion parameters respectively yields a strong edge probability of the pixel to be interpolated.
  3. 根据权利要求2所述的基于边缘自适应的图像缩放方法，其特征在于，所述确定待插值像素的边缘方向包括：The edge-adaptive-based image scaling method according to claim 2, wherein the determining an edge direction of the pixel to be interpolated comprises:
    确定原图像窗口中待插值像素邻域内全部像素梯度加权的协方差矩阵；Determining a covariance matrix of all pixel gradient weights in the neighborhood of the pixel to be interpolated in the original image window;
    计算所述协方差矩阵的特征值和特征向量，并将较小特征值对应的特征向量作为待插值像素的边缘方向；Calculating an eigenvalue and a feature vector of the covariance matrix, and using a feature vector corresponding to the smaller feature value as an edge direction of the pixel to be interpolated;
    相应的，所述根据原图像窗口内全部像素的梯度的协方差矩阵确定边缘方向可靠度参数包括：Correspondingly, determining the edge direction reliability parameter according to the covariance matrix of the gradients of all the pixels in the original image window comprises:
    构造所述协方差矩阵的较大特征值和较小特征值的比值的递增函数 f(x):Constructing an increasing function of a ratio of a larger eigenvalue to a smaller eigenvalue of the covariance matrix f(x):
    Figure PCTCN2015075696-appb-100001
    Figure PCTCN2015075696-appb-100001
    根据公式(11)确定边缘方向可靠度参数Ra:Determine the edge direction reliability parameter Ra according to formula (11):
    Figure PCTCN2015075696-appb-100002
    Figure PCTCN2015075696-appb-100002
    其中，λ1、λ2分别表示所述协方差矩阵的较大特征值和较小特征值，A、B、和C分别表示所述协方差矩阵M的元素(M)11、(M)12或(M)21、和(M)22，T1、T2分别表示边缘可靠度第一阈值和边缘可靠度第二阈值。Where λ 1 and λ 2 respectively represent larger eigenvalues and smaller eigenvalues of the covariance matrix, and A, B, and C represent elements (M) 11 and (M) 12 of the covariance matrix M, respectively. Or (M) 21 and (M) 22 , T1, T2 represent a first threshold of edge reliability and a second threshold of edge reliability, respectively.
  4. 根据权利要求2所述的基于边缘自适应的图像缩放方法，其特征在于，所述根据原图像窗口中待插值像素邻域内全部像素梯度幅值的加权平均确定边缘强度参数包括：The edge-adaptive image scaling method according to claim 2, wherein the determining the edge strength parameter according to the weighted average of all pixel gradient magnitudes in the neighborhood of the pixel to be interpolated in the original image window comprises:
    确定原图像窗口中待插值像素邻域内全部像素梯度幅值的加权平均值AvgG：Determining the weighted average AvgG of all pixel gradient magnitudes in the neighborhood of the pixel to be interpolated in the original image window:
    Figure PCTCN2015075696-appb-100003
    Figure PCTCN2015075696-appb-100003
    根据公式(13)确定边缘强度Rs：Determine the edge strength Rs according to formula (13):
    Figure PCTCN2015075696-appb-100004
    Figure PCTCN2015075696-appb-100004
    其中，w(i,j)表示(i,j)位置像素的双线性插值权重，gH(i,j)表示(i,j)位置的像素的水平梯度，gV(i,j)表示(i,j)位置的像素的垂直梯度，T3、和T4分别表示边缘强度第一阈值和边缘强度第二阈值。Where w(i,j) represents the bilinear interpolation weight of the (i,j) position pixel, g H (i,j) represents the horizontal gradient of the pixel at the (i,j) position, g V (i,j) A vertical gradient representing the pixel at the (i, j) position, T3, and T4 representing the edge intensity first threshold and the edge intensity second threshold, respectively.
  5. 根据权利要求2所述的基于边缘自适应的图像缩放方法，其特征在于，所述根据原图像窗口中待插值像素邻域内全部像素的方向斜率的方差确定边缘方向一致性参数包括：The edge-adapted image scaling method according to claim 2, wherein the determining the edge direction consistency parameter according to the variance of the slope of the direction of all the pixels in the neighborhood of the pixel to be interpolated in the original image window comprises:
    确定插值图像中待插值像素邻域内像素的斜率方差var：Determining the slope variance var of the pixels in the neighborhood of the pixel to be interpolated in the interpolated image:
    Figure PCTCN2015075696-appb-100005
    Figure PCTCN2015075696-appb-100005
    根据公式(15)确定边缘方向一致性参数Rc： Determine the edge direction consistency parameter Rc according to formula (15):
    Figure PCTCN2015075696-appb-100006
    Figure PCTCN2015075696-appb-100006
    其中，
    Figure PCTCN2015075696-appb-100007
    表示待插值像素邻域内像素的方向斜率，
    Figure PCTCN2015075696-appb-100008
    表示待插值像素领域内像素的方向斜率的平均值，T5、和T6分别表示边缘方向一致性第一阈值和边缘方向一致性第二阈值。    among them,
    Figure PCTCN2015075696-appb-100007
     Representing the slope of the direction of the pixel in the neighborhood of the pixel to be interpolated,
    Figure PCTCN2015075696-appb-100008
    The average value of the direction slopes of the pixels in the field of the pixel to be interpolated is represented, and T5, and T6 represent the first threshold value of the edge direction consistency and the second threshold of the edge direction consistency, respectively.
  6. 一种基于边缘自适应的图像缩放系统，其特征在于，包括：An image scaling system based on edge adaptation, comprising:
    坐标计算模块，用于确定待插值像素在原图像中的位置；a coordinate calculation module, configured to determine a position of the pixel to be interpolated in the original image;
    强边检测模块，用于计算待插值像素的边缘方向和强边概率；a strong edge detection module for calculating an edge direction and a strong edge probability of the pixel to be interpolated;
    边缘插值模块，用于在所述待插值像素的强边概率不为零时根据所述边缘方向利用边缘插值法进行插值；An edge interpolation module, configured to perform interpolation by using an edge interpolation method according to the edge direction when a strong edge probability of the pixel to be interpolated is not zero;
    非边缘插值模块，用于在所述待插值像素的强边概率为零时根据非边缘插值法进行插值；a non-edge interpolation module, configured to perform interpolation according to a non-edge interpolation method when a strong edge probability of the pixel to be interpolated is zero;
    融合模块，用于将所述强边概率和所述强边概率的互补概率分别作为边缘插值融合权重和非边缘插值融合权重，对所述边缘插值法和所述非边缘插值法获得的插值结果进行加权融合得到最终的插值后图像。a fusion module, configured to use the strong edge probability and the complementary probability of the strong edge probability as edge interpolation fusion weights and non-edge interpolation fusion weights respectively, and the interpolation results obtained by the edge interpolation method and the non-edge interpolation method A weighted fusion is performed to obtain the final interpolated image.
  7. 根据权利要求6所述的基于边缘自适应的图像缩放系统，其特征在于，所述强边检测模块包括梯度计算模块、边缘方向可靠度子模块、边缘强度子模块、和边缘方向一致性子模块：The edge-adaptive-based image scaling system according to claim 6, wherein the strong edge detection module comprises a gradient calculation module, an edge direction reliability sub-module, an edge intensity sub-module, and an edge direction consistency sub-module:
    所述梯度计算模块，用于计算待插值像素的边缘方向；The gradient calculation module is configured to calculate an edge direction of the pixel to be interpolated;
    所述边缘方向可靠度子模块，用于根据原图像窗口中待插值像素邻域内全部像素梯度加权的协方差矩阵确定边缘方向可靠度参数；The edge direction reliability sub-module is configured to determine an edge direction reliability parameter according to a covariance matrix weighted by all pixel gradients in a neighborhood of the pixel to be interpolated in the original image window;
    所述边缘强度子模块，用于根据原图像窗口中待插值像素邻域内全部像素梯度幅值的加权平均确定边缘强度参数；The edge strength sub-module is configured to determine an edge strength parameter according to a weighted average of all pixel gradient magnitudes in a neighborhood of the pixel to be interpolated in the original image window;
    所述边缘方向一致性子模块，用于根据原图像窗口中待插值像素邻域内全部像素的方向斜率的方差确定边缘方向一致性参数；The edge direction consistency sub-module is configured to determine an edge direction consistency parameter according to a variance of a direction slope of all pixels in a neighborhood of the pixel to be interpolated in the original image window;
    将所确定的所述参数中的一个或其任意组合分别作为所述强边融合参数的相互独立的乘性因子相乘得到待插值像素的强边概率。Multiplying one of the determined parameters or any combination thereof as mutually independent multiplicative factors of the strong edge fusion parameters respectively yields a strong edge probability of the pixel to be interpolated.
  8. 根据权利要求6所述的基于边缘自适应的图像缩放系统，其特征在于，所述梯度计算模块，具体用于确定原图像窗口中待插值像素邻域内 全部像素梯度加权的协方差矩阵、和计算所述协方差矩阵的特征值和特征向量，并将较小特征值对应的特征向量作为待插值像素的边缘方向；The edge-adaptive-based image scaling system according to claim 6, wherein the gradient calculation module is specifically configured to determine a neighborhood of pixels to be interpolated in the original image window. a covariance matrix of all pixel gradient weights, and calculating an eigenvalue and a feature vector of the covariance matrix, and using a feature vector corresponding to the smaller feature value as an edge direction of the pixel to be interpolated;
    所述边缘方向可靠度子模块，具体用于构造所述协方差矩阵的较大特征值和较小特征值的比值的递增函数f(x):The edge direction reliability sub-module is specifically configured to construct an increasing function f(x) of a ratio of a larger eigenvalue and a smaller eigenvalue of the covariance matrix:
    Figure PCTCN2015075696-appb-100009
    Figure PCTCN2015075696-appb-100009
    和根据公式(11)确定边缘方向可靠度参数Ra:And determining the edge direction reliability parameter Ra according to formula (11):
    Figure PCTCN2015075696-appb-100010
    Figure PCTCN2015075696-appb-100010
    其中，λ1、λ2分别表示所述协方差矩阵的较大特征值和较小特征值，A、B、和C分别表示所述协方差矩阵M的元素(M)11、(M)12或(M)21、和(M)22，T1、T2分别表示边缘可靠度第一阈值和边缘可靠度第二阈值。Where λ 1 and λ 2 respectively represent larger eigenvalues and smaller eigenvalues of the covariance matrix, and A, B, and C represent elements (M) 11 and (M) 12 of the covariance matrix M, respectively. Or (M) 21 and (M) 22 , T1, T2 represent a first threshold of edge reliability and a second threshold of edge reliability, respectively.
  9. 根据权利要求6所述的基于边缘自适应的图像缩放系统，其特征在于，所述边缘强度子模块具体用于确定原图像窗口中待插值像素邻域内全部像素梯度幅值的加权平均值AvgG：The edge-adaptive-based image scaling system according to claim 6, wherein the edge intensity sub-module is specifically configured to determine a weighted average value AvgG of all pixel gradient magnitudes in a neighborhood of pixels to be interpolated in the original image window:
    Figure PCTCN2015075696-appb-100011
    Figure PCTCN2015075696-appb-100011
    和根据公式(13)确定边缘强度Rs：And determining the edge strength Rs according to formula (13):
    Figure PCTCN2015075696-appb-100012
    Figure PCTCN2015075696-appb-100012
    其中，w(i,j)表示(i,j)位置像素的双线性插值权重，gH(i,j)表示(i,j)位置的像素的水平梯度，gV(i,j)表示(i,j)位置的像素的垂直梯度，T3、和T4分别表示边缘强度第一阈值和边缘强度第二阈值。Where w(i,j) represents the bilinear interpolation weight of the (i,j) position pixel, g H (i,j) represents the horizontal gradient of the pixel at the (i,j) position, g V (i,j) A vertical gradient representing the pixel at the (i, j) position, T3, and T4 representing the edge intensity first threshold and the edge intensity second threshold, respectively.
  10. 根据权利要求6所述的基于边缘自适应的图像缩放系统，其特征在于，所述边缘方向一致性子模块具体用于确定插值图像中待插值像素邻域内像素的斜率方差var：The edge-adapted image scaling system according to claim 6, wherein the edge direction consistency sub-module is specifically configured to determine a slope variance var of a pixel in a neighborhood of the pixel to be interpolated in the interpolated image:
    Figure PCTCN2015075696-appb-100013
    Figure PCTCN2015075696-appb-100013
    和根据公式(15)确定边缘方向一致性参数Rc： And determining the edge direction consistency parameter Rc according to formula (15):
    Figure PCTCN2015075696-appb-100014
    Figure PCTCN2015075696-appb-100014
    其中，
    Figure PCTCN2015075696-appb-100015
    表示待插值像素邻域内像素的方向斜率，
    Figure PCTCN2015075696-appb-100016
    表示待插值像素领域内像素的方向斜率的平均值，T5、和T6分别表示边缘方向一致性第一阈值和边缘方向一致性第二阈值。     among them,
    Figure PCTCN2015075696-appb-100015
     Representing the slope of the direction of the pixel in the neighborhood of the pixel to be interpolated,
    Figure PCTCN2015075696-appb-100016
    The average value of the direction slopes of the pixels in the field of the pixel to be interpolated is represented, and T5, and T6 represent the first threshold value of the edge direction consistency and the second threshold of the edge direction consistency, respectively.
PCT/CN2015/075696 2015-04-01 2015-04-01 Method and system for image scaling based on edge self-adaptation WO2016154966A1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2015/075696 WO2016154966A1 (en) 2015-04-01 2015-04-01 Method and system for image scaling based on edge self-adaptation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2015/075696 WO2016154966A1 (en) 2015-04-01 2015-04-01 Method and system for image scaling based on edge self-adaptation

Publications (1)

Publication Number Publication Date
WO2016154966A1 true WO2016154966A1 (en) 2016-10-06

Family

ID=57003775

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2015/075696 WO2016154966A1 (en) 2015-04-01 2015-04-01 Method and system for image scaling based on edge self-adaptation

Country Status (1)

Country Link
WO (1) WO2016154966A1 (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20040091174A1 (en) * 2003-10-30 2004-05-13 Samsung Electronics Co., Ltd. Method and apparatus for image interpolation based on adaptive polyphase filters
CN1667650A (en) * 2005-04-08 2005-09-14 杭州国芯科技有限公司 Image zooming method based on edge detection
CN102567950A (en) * 2011-12-27 2012-07-11 深圳市万兴软件有限公司 Image scaling method and system
CN104200426A (en) * 2014-08-25 2014-12-10 北京京东方视讯科技有限公司 Image interpolation method and device
CN104700360A (en) * 2015-04-01 2015-06-10 中国科学院自动化研究所 Image zooming method and system based on edge self-adaptation

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20040091174A1 (en) * 2003-10-30 2004-05-13 Samsung Electronics Co., Ltd. Method and apparatus for image interpolation based on adaptive polyphase filters
CN1667650A (en) * 2005-04-08 2005-09-14 杭州国芯科技有限公司 Image zooming method based on edge detection
CN102567950A (en) * 2011-12-27 2012-07-11 深圳市万兴软件有限公司 Image scaling method and system
CN104200426A (en) * 2014-08-25 2014-12-10 北京京东方视讯科技有限公司 Image interpolation method and device
CN104700360A (en) * 2015-04-01 2015-06-10 中国科学院自动化研究所 Image zooming method and system based on edge self-adaptation

Similar Documents

Publication Publication Date Title
CN104700360B (en) Image-scaling method and system based on edge self-adaption
Liu et al. Image interpolation via regularized local linear regression
Hung et al. Fast image interpolation using the bilateral filter
JP5822157B2 (en) Noise reduction apparatus, noise reduction method, and program
TWI543586B (en) Method for image enhancement, and image processing apparatus and computer program product using the same
US11145032B2 (en) Image processing apparatus, method and storage medium for reducing color noise and false color
TWI546777B (en) Image processing apparatus and method
Jeong et al. Multi-frame example-based super-resolution using locally directional self-similarity
WO2015198368A1 (en) Image processing device and image processing method
US9508025B2 (en) Image processing device, image processing method and medium
Lu et al. Video super resolution based on non-local regularization and reliable motion estimation
Singh et al. Content adaptive single image interpolation based Super Resolution of compressed images
JP4868249B2 (en) Video signal processing device
Mishiba et al. Edge-adaptive image interpolation using constrained least squares
WO2016154966A1 (en) Method and system for image scaling based on edge self-adaptation
JP6938282B2 (en) Image processing equipment, image processing methods and programs
Park et al. Covariance-based adaptive deinterlacing method using edge map
US9401005B2 (en) Image processing circuit, image processing method, and display device using the same
CN105069764A (en) Image denoising method and system based on edge tracking
JP2009157449A (en) Image processing system, image processing method, and program for image processing
JP6957665B2 (en) Image processing equipment, image processing methods and programs
JP5705391B1 (en) Image processing apparatus and image processing method
Al-Mokhtar et al. A review of digital image fusion and its application
WO2016154970A1 (en) Method and system for image interpolation based on edge detection
WO2014136552A1 (en) Image processing device

Legal Events

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

Ref document number: 15886936

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: 15886936

Country of ref document: EP

Kind code of ref document: A1