WO2016154970A1 - 基于边缘检测的图像插值方法及系统 - Google Patents

基于边缘检测的图像插值方法及系统 Download PDF

Info

Publication number
WO2016154970A1
WO2016154970A1 PCT/CN2015/075709 CN2015075709W WO2016154970A1 WO 2016154970 A1 WO2016154970 A1 WO 2016154970A1 CN 2015075709 W CN2015075709 W CN 2015075709W WO 2016154970 A1 WO2016154970 A1 WO 2016154970A1
Authority
WO
WIPO (PCT)
Prior art keywords
interpolated
pixel
intersection
interpolation
original image
Prior art date
Application number
PCT/CN2015/075709
Other languages
English (en)
French (fr)
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/CN2015/075709 priority Critical patent/WO2016154970A1/zh
Publication of WO2016154970A1 publication Critical patent/WO2016154970A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR 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

Definitions

  • the invention belongs to the field of image processing, and in particular relates to an image interpolation method and system based on edge detection.
  • Image interpolation can be used for image resolution adjustment, such as magnifying high-definition images (1920*1080) into ultra-high definition images (3840*2160).
  • a more advanced image interpolation method is an image interpolation method based on edge detection. Through edge detection, the edge direction of the pixel to be interpolated is calculated, and the interpolated pixels are interpolated along the edge direction, thereby obtaining a smooth image edge and avoiding aliasing.
  • the existing image interpolation method based on edge detection has at least one of the following disadvantages: only integer multiple image enlargement is supported; the edge direction used in interpolation is not arbitrary direction, only a few fixed directions; for interpolation The number of original points is small, causing the edges of the image to be unclear.
  • the invention provides an image interpolation method and system based on edge detection, so as to achieve the purpose of making the edge of the image after interpolation and clear and non-aliased.
  • a first aspect of the present invention provides an image interpolation method based on edge detection, including:
  • the interpolation is performed according to the line intersection method and/or the column intersection method, and the line intersection method and/or the column intersection method include:
  • a second aspect of the present invention provides an image interpolation system based on edge detection, including:
  • a coordinate calculation unit configured to determine a position of the pixel to be interpolated in the original image according to the original image and the size of the interpolated image
  • a direction calculating unit configured to determine an edge direction of the pixel to be interpolated in the original image
  • An intersection calculation unit configured to calculate, when the absolute value of the slope of the edge direction is not less than the first threshold, calculate a number of rows and/or columns in the neighborhood of the pixel to be interpolated in the original image, which are determined by a line to be interpolated and an edge direction The location of a number of intersections and/or intersections of several columns;
  • An edge interpolation filtering unit is configured to perform interpolation according to a line intersection method and/or a column intersection method, and is specifically configured to determine the line by using a one-dimensional interpolation method according to values of pixels in a neighborhood of the row intersection and/or a column intersection point in the original image. Pixel values of intersection points and/or column intersection points, and pixel values of row intersection points and/or column intersection points in the determined neighborhood of the pixel to be interpolated are one-dimensionally filtered, the values of the pixels to be interpolated are obtained, and the original image is interpolated.
  • the image interpolation method based on edge detection 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 edges of the interpolated image are clear and the sawtooth phenomenon is avoided.
  • Embodiment 1 is a flowchart of Embodiment 1 of an image interpolation method based on edge detection according to the present invention
  • FIG. 2 is a schematic diagram of a Sobel gradient method in Embodiment 1 of an image interpolation method based on edge detection according to the present invention
  • FIG. 3 is a schematic diagram of a gradient covariance matrix method in Embodiment 1 of an image interpolation method based on edge detection according to the present invention
  • FIG. 4 is a schematic diagram of a line intersection method in Embodiment 1 of an image interpolation method based on edge detection according to the present invention
  • FIG. 5 is a schematic diagram of a column intersection method in Embodiment 1 of an image interpolation method based on edge detection according to the present invention
  • 6 is a weight function in the first embodiment of the image interpolation method based on edge detection according to the present invention, in combination with the application of the intersection method and the column intersection method;
  • FIG. 7 is a structural block diagram of Embodiment 1 of an image interpolation system based on edge detection according to the present invention.
  • FIG. 1 is a flowchart of Embodiment 1 of an image interpolation method based on edge detection according to the present invention.
  • the image interpolation method based on edge detection according to the present invention includes:
  • the determining, according to the size of the original image and the interpolated image, the position of the pixel to be interpolated in the original image includes: Calculate the position of the pixel to be interpolated in the original image according to formula (1):
  • 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 first method for determining an edge direction of the pixel to be interpolated in the original image may include:
  • FIG. 2 is a schematic diagram of a Sobel gradient method in the first embodiment of the image interpolation method based on edge detection according to the present invention.
  • the Sobel gradient operator is used to calculate the to-be-interpolated values in the original image according to formulas (2) and (3), respectively.
  • the edge direction of the pixel to be interpolated is the vertical direction of the direction of the gradient of the pixel to be interpolated (g V (i L , j L ) , -g H (i L ,j L )):
  • g H (i L ,j L ) (1-dx)*(1-dy)*g H (i,j)+dx*(1-dy)*g H (i,j+1)
  • g V (i L ,j L ) (1-dx)*(1-dy)*g V (i,j)+dx*(1-dy)*g V (i,j+1)
  • I L (i-1, j+1), I L (i, j+1), I L (i+1, j+1), I L (i-1, j-1), I L (i, j-1), I L (i+1, j-1), I L (i-1, j), I L (i+1, j) respectively represent eight pixels in the neighborhood of the pixel to be interpolated in the original image Pixel values of pixels;
  • FIG. 3 is a schematic diagram of a gradient covariance matrix method in the first embodiment of the image interpolation method based on the edge detection according to the present invention.
  • the second method for determining the edge direction of the pixel to be interpolated in the original image is shown in FIG. Can include:
  • a feature vector corresponding to a smaller eigenvalue of the covariance matrix v x represents a horizontal component of the edge direction, and v y represents a vertical component of the edge direction.
  • the third method for determining the edge direction of the pixel to be interpolated in the original image may include the steps described in the second method and further includes:
  • the covariance matrix is improved according to formula (8) to obtain an improved covariance matrix M':
  • w(i,j) adopts the method of bilinear interpolation, which is:
  • the interpolation is performed according to the line intersection method and/or the column intersection method, that is, the edge interpolation method.
  • the line intersection method and/or the column intersection method include:
  • S131 Determine an absolute value of a slope of an edge direction. If the value is smaller than a second threshold T1, perform interpolation according to a line intersection method.
  • the interpolation is performed according to the line intersection method and the column intersection method, including:
  • S1331 Calculating a position of a plurality of rows and/or a plurality of column intersections of a plurality of rows and/or columns in the neighborhood of the pixel to be interpolated in the original image to be interpolated by the pixel to be interpolated and the edge direction, including:
  • FIG. 4 is a schematic diagram of a line intersection method in the first embodiment of the image interpolation method based on edge detection according to the present invention.
  • the line intersection method is implemented by using an intersection point of four lines of pixels to be interpolated;
  • the calculating, in the original image, a plurality of rows in the neighborhood of the pixel to be interpolated are to be interpolated pixels and
  • the positions of the intersections of the lines intersected by the straight line determined by the edge direction include calculating the positions of the intersections of the four lines according to the formulas (9), (10), (11), and (12), respectively:
  • FIG. 5 is a schematic diagram of a column intersection method in the first embodiment of the image interpolation method based on the edge detection according to the present invention.
  • the column intersection method uses the intersection of the left and right columns of the pixel to be interpolated, correspondingly,
  • the calculating the positions of the intersections of the plurality of columns of the plurality of columns in the neighborhood of the pixel to be interpolated and the direction determined by the edge direction in the neighborhood of the pixel to be interpolated includes according to formulas (13), (14), (15), and (16) Calculate the position of the intersection of the four columns separately:
  • S1332 Determine, by using a one-dimensional interpolation method, pixel values of the line intersection point and/or the column intersection point according to the value of the pixel in the neighborhood of the row intersection and/or the column intersection point in the original image, including:
  • the column intersection method is implemented by using the intersection of the left and right columns of the pixel to be interpolated.
  • the one-dimensional interpolation method is used according to the original image.
  • the value of the pixel in the neighborhood of the column intersection determines the pixel value of the column intersection including determining the pixel value of the column intersection according to formula (21):
  • Performing one-dimensional filtering on the pixel values of the row intersections and/or column intersections in the determined neighborhood of the pixel to be interpolated, and obtaining the value of the pixel to be interpolated includes performing one-dimensional filtering according to formula (22) to obtain the value of the pixel 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)
  • (i L , j L ) represents the coordinates of the position of the pixel to be interpolated in the original image
  • i and j represent the number of rows and columns, respectively
  • (v x , v y ) represents the edge direction P 0 , P 1 , P 2 , and P 3 respectively represent intersections of four lines
  • I P0 , I P1 , I P2 , and I P3 respectively represent pixel values of intersections of four lines
  • [f 0 , f 1 , f 2 , f 3 ] is the coefficient of the one-dimensional filter, such as [1,3,3,1];
  • the edge direction of the pixel to be interpolated and several columns in the neighborhood are calculated. If the edge direction is the vertical direction, it has no intersection with several columns in the neighborhood of the pixel to be interpolated; when the absolute value of the edge direction slope is larger than the set first threshold k T2 , it is the pixel to be interpolated The intersections of several columns in the neighborhood are far away, and the correlation with the pixels to be interpolated is relatively small; therefore, non-edge maps are used for the above two cases.
  • Interpolation method for interpolating the interpolated two-dimensional image is decomposed into two one-dimensional horizontal and vertical directions sequentially order interpolation, the interpolation in the horizontal direction and the vertical direction interpolation exchangeable.
  • S1333 Perform one-dimensional filtering on pixel values of row intersections and/or column intersections in the neighborhood of the pixels to be interpolated, obtain values of pixels to be interpolated, and interpolate the original image, including:
  • FIG. 6 is a weight function when the line intersection method and the column intersection method are applied in the first embodiment of the image interpolation method based on edge detection according to the present invention, that is, when the weights are generated by the curve shown in FIG. Weight, and weighting the value of the pixel to be interpolated I H (i H ,j H ) according to formula (23):
  • I H (i H ,j H ) w*I HR (i H ,j H )+(1-w)*I HC (i H ,j H ) (23)
  • the interpolation method is used for interpolation; in other directions, the intersection method is used for interpolation, T1 and T2 are preset threshold values, and the method of combining the intersection point method and the column intersection method It is not limited to the form described above;
  • the edge detection based image interpolation method further comprises:
  • the interpolation is performed according to the non-edge interpolation method; that is, 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, then If the pixels to be interpolated have no direction, the non-edge image interpolation method is used for interpolation, and the two-dimensional image interpolation is decomposed into horizontal and vertical two-dimensional directions for interpolation, and the order of horizontal interpolation and vertical interpolation can be exchanged.
  • the image interpolation method based on edge detection 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 edges of the interpolated image are clear and the sawtooth phenomenon is avoided.
  • FIG. 7 is a structural block diagram of Embodiment 1 of an image interpolation system based on edge detection according to the present invention.
  • the image interpolation system based on edge detection of the present invention includes:
  • a coordinate calculation unit configured to determine a position of the pixel to be interpolated in the original image according to the original image and the size of the interpolated image
  • a direction calculating unit configured to determine an edge direction of the pixel to be interpolated in the original image
  • An intersection calculation unit configured to calculate, when the absolute value of the slope of the edge direction is not less than the first threshold, calculate a number of rows and/or columns in the neighborhood of the pixel to be interpolated in the original image, which are determined by a line to be interpolated and an edge direction The location of a number of intersections and/or intersections of several columns;
  • An edge interpolation filtering unit is configured to perform interpolation according to a line intersection method and/or a column intersection method, and is specifically configured to determine the line by using a one-dimensional interpolation method according to values of pixels in a neighborhood of the row intersection and/or a column intersection point in the original image. Pixel values of intersection points and/or column intersection points, and pixel values of row intersection points and/or column intersection points in the determined neighborhood of the pixel to be interpolated are one-dimensionally filtered, the values of the pixels to be interpolated are obtained, and the original image is interpolated.
  • the edge detection based image interpolation system further comprises:
  • a non-edge interpolation unit configured to perform interpolation according to a non-edge interpolation method when an absolute value of a slope in an edge direction is less than a set threshold
  • a fusion unit configured to fuse the result obtained by interpolating the row intersection method and/or the column intersection method with the result obtained by the non-edge interpolation method to obtain an interpolated image.
  • the direction calculation unit is specifically configured to calculate a horizontal gradient g H (i, j) and a vertical gradient g V (i) of a plurality of pixels in the neighborhood of the pixel to be interpolated in the original image according to formulas (2) and (3), respectively.
  • g H horizontal gradient
  • g V vertical gradient
  • g H (i L ,j L ) (1-dx)*(1-dy)*g H (i,j)+dx*(1-dy)*g H (i,j+1)
  • g V (i L ,j L ) (1-dx)*(1-dy)*g V (i,j)+dx*(1-dy)*g V (i,j+1)
  • I L (i-1, j+1), I L (i, j+1), I L (i+1, j+1), I L (i-1, j-1), I L (i, j-1), I L (i+1, j-1), I L (i-1, j), I L (i+1, j) respectively represent eight pixels in the neighborhood of the pixel to be interpolated in the original image The pixel value of the pixel.

Landscapes

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

Abstract

本发明提供一种基于边缘检测的图像插值方法及系统，所述图像插值方法包括：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；确定待插值像素在原图像中的边缘方向；若边缘方向的斜率的绝对值不小于第一阈值，则根据行交点法和/或列交点法进行插值。本发明所述插值方法能够使插值后图像的边缘清晰且无锯齿现象。

Description

基于边缘检测的图像插值方法及系统 技术领域
本发明属于图像处理领域，尤其涉及基于边缘检测的图像插值方法及系统。
背景技术
图像插值，可用于图像的分辨率调整，如将高清图像(1920*1080)放大为超高清图像(3840*2160)。
传统的图像插值方法，如双线性插值，双立方插值，多相位插值等，本质是使用低通滤波器进行插值，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。目前，较先进的一种图像插值方法为基于边缘检测的图像插值方法。通过边缘检测，计算出待插值像素的边缘方向，沿边缘方向对待插值像素进行插值，从而获得平滑的图像边缘，避免锯齿现象。但是，现有的基于边缘检测的图像插值方法至少存在着下述缺点之一：只支持整数倍图像放大；插值时使用的边缘方向不是任意方向，只有少数的固定的几个方向；用于插值的原始点数量较少，造成图像的边缘不够清晰。
因此，需要一种能够解决上述问题的图像插值方法。
发明内容
本发明提供一种基于边缘检测的图像插值方法及系统，以实现使插值后图像边缘清晰且无锯齿的目的。
本发明的第一个方面是提供一种基于边缘检测的图像插值方法，包括：
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；
确定待插值像素在原图像中的边缘方向；
若边缘方向的斜率的绝对值不小于第一阈值，则根据行交点法和/或列交点法进行插值，所述行交点法和/或列交点法包括：
计算原图像中待插值像素邻域内若干行和/或若干列被待插值像素和边缘方向确定的直线所截的若干行交点和/或若干列交点的位置；
利用一维插值法根据原图像中所述行交点和/或列交点邻域内像素的值确定所述行交点和/或列交点的像素值；
对所确定的待插值像素邻域内的行交点和/或列交点的像素值进行一维滤波，获得待插值像素的值，并对原图像进行插值。
本发明的第二个方面是提供一种基于边缘检测的图像插值系统，包括：
坐标计算单元，用于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；
方向计算单元，用于确定待插值像素在原图像中的边缘方向；
交点计算单元，用于在边缘方向的斜率的绝对值不小于第一阈值时，计算原图像中待插值像素邻域内若干行和/或若干列被待插值像素和边缘方向确定的直线所截的若干行交点和/或若干列交点的位置；
边缘插值滤波单元，用于根据行交点法和/或列交点法进行插值，具体用于利用一维插值法根据原图像中所述行交点和/或列交点邻域内像素的值确定所述行交点和/或列交点的像素值、和对所确定的待插值像素邻域内的行交点和/或列交点的像素值进行一维滤波，获得待插值像素的值，并对原图像进行插值。
本发明的有益效果为：
本发明基于边缘检测的图像插值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，使插值后的图像边缘清晰且避免了锯齿现象。
附图说明
图1为本发明基于边缘检测的图像插值方法实施例一的流程图；
图2为本发明基于边缘检测的图像插值方法实施例一中的Sobel梯度方法示意图；
图3为本发明基于边缘检测的图像插值方法实施例一中的梯度协方差矩阵方法示意图；
图4为本发明基于边缘检测的图像插值方法实施例一中的行交点法示意图；
图5为本发明基于边缘检测的图像插值方法实施例一中的列交点法示意图；
图6为本发明基于边缘检测的图像插值方法实施例一中结合应用行交点法和列交点法时的权重函数；
图7为本发明基于边缘检测的图像插值系统实施例一的结构框图。
具体实施方式
图1为本发明基于边缘检测的图像插值方法实施例一的流程图，如图1所示，本发明基于边缘检测的图像插值方法，包括：
S11、根据原图像和插值后图像的大小也即分辨率、确定待插值像素在原图像中的位置；优选的，所述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根据公式(1)计算待插值像素在原图像中的位置：
Figure PCTCN2015075709-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；
S12、确定待插值像素在原图像中的边缘方向；优选的，所述确定待插值像素在原图像中的边缘方向的第一种方法可以包括：
图2为本发明基于边缘检测的图像插值方法实施例一中的Sobel梯度方法示意图，如图2所示，采用Sobel梯度算子，分别根据公式(2)和(3)计算原图像中待插值像素邻域内若干像素的水平梯度gH(i,j)和垂直梯度gV(i,j)：
gH(i,j)＝IL(i-1,j+1)+2*IL(i,j+1)+IL(i+1,j+1)
                             (2)
-IL(i-1,j-1)-2*IL(i,j-1)-IL(i+1,j-1)
gV(i,j)＝IL(i-1,j-1)+2*IL(i-1,j)+IL(i-1,j+1)
                            (3)
-IL(i+1,j-1)-2*IL(i+1,j)-IL(i+1,j+1)
分别根据所述待插值像素在原图像中的位置与所述邻域内像素的水 平梯度和垂直梯度利用双线性插值也即根据公式(4)和(5)确定待插值像素的水平梯度gH(iL,jL)和垂直梯度gV(iL,jL)，则待插值像素的边缘方向即是所述待插值像素的梯度的方向的垂直方向(gV(iL,jL),-gH(iL,jL))：
gH(iL,jL)＝(1-dx)*(1-dy)*gH(i,j)+dx*(1-dy)*gH(i,j+1)
                        (4)
+(1-dx)*dy*gH(i+1,j)+dx*dy*gH(i+1,j+1)
gV(iL,jL)＝(1-dx)*(1-dy)*gV(i,j)+dx*(1-dy)*gV(i,j+1)
                       (5)
+(1-dx)*dy*gV(i+1,j)+dx*dy*gV(i+1,j+1)
其中，IL(i-1,j+1)、IL(i,j+1)、IL(i+1,j+1)、IL(i-1,j-1)、IL(i,j-1)、IL(i+1,j-1)、IL(i-1,j)、IL(i+1,j)分别表示在原图像中待插值像素邻域内八个像素的像素值；
其次的，图3为本发明基于边缘检测的图像插值方法实施例一中的梯度协方差矩阵方法示意图，如图3所示，所述确定待插值像素在原图像中的边缘方向的第二种方法可以包括：
选取待插值像素邻域内H*W的窗口Ω任意大小的窗口，例如本例中H＝4，W＝6；确定窗口内全部像素的水平梯度gH(i,j)和垂直梯度gV(i,j)，从而确定待插值像素邻域内窗口中全部像素的协方差矩阵M：
Figure PCTCN2015075709-appb-000002
计算所述协方差矩阵的特征值和特征向量，则确定较小特征值对应的特征向量v为所述边缘方向，也即：
Figure PCTCN2015075709-appb-000003
其中，
Figure PCTCN2015075709-appb-000004
表示所述协方差矩阵的较小特征值对应的特征向量；vx表示边缘方向的水平分量，vy表示边缘方向的垂直分量。
此外，所述确定待插值像素在原图像中的边缘方向第三种方法可以包括第二种方法所述各步骤以及还包括：
根据公式(8)改进所述协方差矩阵获得改进后的协方差矩阵M'：
Figure PCTCN2015075709-appb-000005
其中，w(i,j)的取值采用双线性插值的取法，即为：
w(i-1,j-2)＝(1-dx)*(1-dy)，w(i-1,j-1)＝(1-dy)，w(i-1,j)＝(1-dy)，
w(i-1,j+1)＝(1-dy)，w(i-1,j+2)＝(1-dy)，w(i-1,j+3)＝dx*(1-dy)；
w(i,j-2)＝(1-dx)，w(i,j-1)＝1，w(i,j)＝1，w(i,j+1)＝1，w(i,j+2)＝1，
w(i,j+3)＝dx；w(i+1,j-2)＝(1-dx)，w(i+1,j-1)＝1，w(i+1,j)＝1，w(i+1,j+1)＝1，
w(i+1,j+2)＝1，w(i+1,j+3)＝dx；w(i+2,j-2)＝(1-dx)*dy，w(i+2,j-1)＝dy，
w(i+2,j)＝dy，w(i+2,j+1)＝dy，w(i+2,j+2)＝dy，w(i+2,j+3)＝dx*dy；w(i,j)取值也可以用表1表示：
(1-dx)*(1-dy) (1-dy) (1-dy) (1-dy) (1-dy) dx*(1-dy)
(1-dx) 1 1 1 1 dx
(1-dx) 1 1 1 1 dx
(1-dx)*dy dy dy dy dy dx*dy
表1
S13、若边缘方向的斜率的绝对值不小于第一阈值，则根据行交点法和/或列交点法、也即边缘插值法进行插值，优选的，所述行交点法和/或列交点法包括：
S131、判断边缘方向的斜率的绝对值，若小于第二阈值T1，则根据行交点法进行插值；
S132、若不小于第三阈值T2，则根据列交点法进行插值；
S133、若不小于第二阈值T1而小于第三阈值T2，则同时根据行交点法和列交点法进行插值，包括：
S1331、计算原图像中待插值像素邻域内若干行和/或若干列被待插值像素和边缘方向确定的直线所截的若干行交点和/或若干列交点的位置，包括：
计算原图像中待插值像素邻域内若干行和若干列被待插值像素和边缘方向确定的直线所截的若干行交点和若干列交点的位置；
优选的，图4为本发明基于边缘检测的图像插值方法实施例一中的行交点法示意图，如图4所示，所述行交点法采用待插值像素上下四行的交点实现；
相应的，所述计算原图像中待插值像素邻域内若干行被待插值像素和 边缘方向确定的直线所截的若干行交点的位置包括根据公式(9)、(10)、(11)、和(12)分别计算四个行交点的位置：
Figure PCTCN2015075709-appb-000006
Figure PCTCN2015075709-appb-000007
Figure PCTCN2015075709-appb-000008
Figure PCTCN2015075709-appb-000009
同理，图5为本发明基于边缘检测的图像插值方法实施例一中的列交点法示意图，如图5所示，所述列交点法采用待插值像素左右四列的交点实现，相应的，所述计算原图像中待插值像素邻域内若干列被待插值像素和边缘方向确定的直线所截的若干列交点的位置包括根据公式(13)、(14)、(15)、和(16)分别计算四个列交点的位置：
Figure PCTCN2015075709-appb-000010
Figure PCTCN2015075709-appb-000011
Figure PCTCN2015075709-appb-000012
Figure PCTCN2015075709-appb-000013
S1332、利用一维插值法根据原图像中所述行交点和/或列交点邻域内像素的值确定所述行交点和/或列交点的像素值，包括：
利用一维插值法根据原图像中所述行交点和列交点邻域内像素的值确定所述行交点和列交点的像素值，优选的，包括根据公式(17)、(18)、(19)、和(20)确定四个行交点的像素值：
Figure PCTCN2015075709-appb-000014
Figure PCTCN2015075709-appb-000015
Figure PCTCN2015075709-appb-000016
Figure PCTCN2015075709-appb-000017
同理，所述列交点法采用待插值像素左右四列的交点实现，相应的，仅举第一个列交点的像素值的计算为例，所述利用一维插值法根据原图像中所述列交点邻域内像素的值确定所述列交点的像素值包括根据公式(21)确定列交点的像素值：
Figure PCTCN2015075709-appb-000018
其他三个列交点的像素值的计算不再赘述；
所述对所确定的待插值像素邻域内的行交点和/或列交点的像素值进行一维滤波，获得待插值像素的值包括根据公式(22)进行一维滤波获得待插值像素的值：
IH(iH,jH)＝f0*IP0+f1*IP1+f2*IP2+f3*IP3   (22)
其中，[]表示向下取整，(iL,jL)表示待插值像素在原图像中的位置的坐标，i和j分别表示行数和列数，(vx,vy)表示边缘方向，P0、P1、P2、P3分别表示四个行交点，IP0、IP1、IP2、IP3分别表示四个行交点的像素值， [f0,f1,f2,f3]为一维滤波器的系数，如[1，3，3，1]；
需要说明的是，计算待插值像素的边缘方向与其邻域内若干行的交点时，若边缘方向为水平方向，则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，小于设定的第一阈值kT1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；同理，计算待插值像素的边缘方向与其邻域内若干列的交点时，若边缘方向为垂直方向，则其与待插值像素邻域内若干列没有交点；当边缘方向斜率的绝对值较大，大于设定的第一阈值kT2时，其与待插值像素邻域内若干列的交点较远，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。
S1333、对所确定的待插值像素邻域内的行交点和/或列交点的像素值进行一维滤波，获得待插值像素的值，并对原图像进行插值，包括：
对所确定的待插值像素邻域内的行交点和列交点的像素值分别进行一维滤波得到行交点滤波的插值结果IHR(iH,jH)和列交点滤波的插值结果IHC(iH,jH)，图6为本发明基于边缘检测的图像插值方法实施例一中结合应用行交点法和列交点法时的权重函数，也即通过图6所示的曲线生成二者加权时的权重，并根据公式(23)加权确定待插值像素的值IH(iH,jH)：
IH(iH,jH)＝w*IHR(iH,jH)+(1-w)*IHC(iH,jH)       (23)
再根据所述待插值像素的值对原图像进行插值；
其中，(iH,jH)表示待插值像素的位置的坐标，w表示行交点滤波的插值结果的权重；
需要说明的是，在低角度时，采用列交点的方法进行插值；在其他方向，采用行交点的方法进行插值，T1和T2为预先设定的阈值，行交点方法和列交点方法的结合方法并不限于以上描述的形式；
优选的，所述基于边缘检测的图像插值方法还包括：
S14、若边缘方向的斜率的绝对值小于设定阈值，则根据非边缘插值 法进行插值；也就是说，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，则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。
S15、对所述行交点法和/或列交点法插值得到的结果与所述非边缘插法值得到的结果进行融合从而获得插值后的图像。
本发明基于边缘检测的图像插值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，使插值后的图像边缘清晰且避免了锯齿现象。
图7为本发明基于边缘检测的图像插值系统实施例一的结构框图，如图7所示，本发明基于边缘检测的图像插值系统包括：
坐标计算单元，用于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；
方向计算单元，用于确定待插值像素在原图像中的边缘方向；
交点计算单元，用于在边缘方向的斜率的绝对值不小于第一阈值时，计算原图像中待插值像素邻域内若干行和/或若干列被待插值像素和边缘方向确定的直线所截的若干行交点和/或若干列交点的位置；
边缘插值滤波单元，用于根据行交点法和/或列交点法进行插值，具体用于利用一维插值法根据原图像中所述行交点和/或列交点邻域内像素的值确定所述行交点和/或列交点的像素值、和对所确定的待插值像素邻域内的行交点和/或列交点的像素值进行一维滤波，获得待插值像素的值，并对原图像进行插值。
优选的，所述基于边缘检测的图像插值系统还包括：
非边缘插值单元，用于在边缘方向的斜率的绝对值小于设定阈值时，根据非边缘插值法进行插值；
融合单元，用于对所述行交点法和/或列交点法插值得到的结果与所述非边缘插值法得到的结果进行融合从而获得插值后的图像。
优选的，所述方向计算单元具体用于：分别根据公式(2)和(3)计算原图像中待插值像素邻域内若干像素的水平梯度gH(i,j)和垂直梯度gV(i,j)：
gH(i,j)＝IL(i-1,j+1)+2*IL(i,j+1)+IL(i+1,j+1)
                                 (2)
-IL(i-1,j-1)-2*IL(i,j-1)-IL(i+1,j-1)
gV(i,j)＝IL(i-1,j-1)+2*IL(i-1,j)+IL(i-1,j+1)
                                 (3)
-IL(i+1,j-1)-2*IL(i+1,j)-IL(i+1,j+1)
和分别根据所述待插值像素在原图像中的位置与所述邻域内像素的水平梯度和垂直梯度利用双线性插值也即根据公式(4)和(5)确定待插值像素的水平梯度gH(iL,jL)和垂直梯度gV(iL,jL)，则待插值像素的边缘方向即是所述待插值像素的梯度的方向的垂直方向(gV(iL,jL),-gH(iL,jL))：
gH(iL,jL)＝(1-dx)*(1-dy)*gH(i,j)+dx*(1-dy)*gH(i,j+1)
                                (4)
+(1-dx)*dy*gH(i+1,j)+dx*dy*gH(i+1,j+1)
gV(iL,jL)＝(1-dx)*(1-dy)*gV(i,j)+dx*(1-dy)*gV(i,j+1)
                             (5)
+(1-dx)*dy*gV(i+1,j)+dx*dy*gV(i+1,j+1)
其中，IL(i-1,j+1)、IL(i,j+1)、IL(i+1,j+1)、IL(i-1,j-1)、IL(i,j-1)、IL(i+1,j-1)、IL(i-1,j)、IL(i+1,j)分别表示在原图像中待插值像素邻域内八个像素的像素值。
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。

Claims (10)

  1. 一种基于边缘检测的图像插值方法，其特征在于，包括：
    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；
    确定待插值像素在原图像中的边缘方向；
    若边缘方向的斜率的绝对值不小于第一阈值，则根据行交点法和/或列交点法进行插值，所述行交点法和/或列交点法包括：
    计算原图像中待插值像素邻域内若干行和/或若干列被待插值像素和边缘方向确定的直线所截的若干行交点和/或若干列交点的位置；
    利用一维插值法根据原图像中所述行交点和/或列交点邻域内像素的值确定所述行交点和/或列交点的像素值；
    对所确定的待插值像素邻域内的行交点和/或列交点的像素值进行一维滤波，获得待插值像素的值，并对原图像进行插值。
  2. 根据权利要求1所述的基于边缘检测的图像插值方法，其特征在于，所述行交点法和/或列交点法包括：
    判断边缘方向的斜率的绝对值，若小于第二阈值T1，则根据行交点法进行插值；
    若不小于第三阈值T2，则根据列交点法进行插值；
    若不小于第二阈值T1而小于第三阈值T2，则同时根据行交点法和列交点法进行插值，包括：
    计算原图像中待插值像素邻域内若干行和若干列被待插值像素和边缘方向确定的直线所截的若干行交点和若干列交点的位置；
    利用一维插值法根据原图像中所述行交点和列交点邻域内像素的值确定所述行交点和列交点的像素值；
    对所确定的待插值像素邻域内的行交点和列交点的像素值分别进行一维滤波得到行交点滤波的插值结果IHR(iH,jH)和列交点滤波的插值结果IHC(iH,jH)，根据公式(23)加权确定待插值像素的值IH(iH,jH)：
    IH(iH,jH)＝w*IHR(iH,jH)+(1-w)*IHC(iH,jH)  (23)
    再根据所述待插值像素的值对原图像进行插值；
    其中，(iH,jH)表示待插值像素的位置的坐标，w表示行交点滤波的插值结果的权重。
  3. 根据权利要求1所述的基于边缘检测的图像插值方法，其特征在于，还包括：
    若边缘方向的斜率的绝对值小于设定阈值，则根据非边缘插值法进行插值；
    相应的，在所述对所确定的待插值像素邻域内的行交点和/或列交点的像素值进行一维滤波，获得待插值像素的值，并对原图像进行插值之后且在所述根据非边缘插值法进行插值之后，还包括：
    对所述行交点法和/或列交点法插值得到的结果与所述非边缘插值法得到的结果进行融合从而获得插值后的图像。
  4. 根据权利要求1所述的基于边缘检测的图像插值方法，其特征在于，所述行交点法采用待插值像素上下四行的交点实现；
    相应的，所述计算原图像中待插值像素邻域内若干行被待插值像素和边缘方向确定的直线所截的若干行交点的位置包括根据公式(9)、(10)、(11)、和(12)分别计算四个行交点的位置：
    Figure PCTCN2015075709-appb-100001
    Figure PCTCN2015075709-appb-100002
    Figure PCTCN2015075709-appb-100003
    Figure PCTCN2015075709-appb-100004
    所述利用一维插值法根据原图像中所述行交点邻域内像素的值确定所述行交点的像素值包括根据公式(17)、(18)、(19)、和(20)确定四个行交点的像素值：
    Figure PCTCN2015075709-appb-100005
    Figure PCTCN2015075709-appb-100006
    Figure PCTCN2015075709-appb-100007
    Figure PCTCN2015075709-appb-100008
    所述对所确定的待插值像素邻域内的行交点的像素值进行一维滤波，获得待插值像素的值包括根据公式(22)进行一维滤波获得待插值像素的值：
    IH(iH,jH)＝f0*IP0+f1*IP1+f2*IP2+f3*IP3  (22)
    其中，[]表示向下取整，(iL,jL)表示待插值像素在原图像中的位置的坐标，i和j分别表示行数和列数，(vx,vy)表示边缘方向，vx表示边缘方向的水平分量，vy表示边缘方向的垂直分量，P0、P1、P2、P3分别表示四个行交点，IP0、IP1、IP2、IP3分别表示四个行交点的像素值，[f0,f1,f2,f3]为一维滤波器的系数。
  5. 根据权利要求1所述的基于边缘检测的图像插值方法，其特征在于，所述确定待插值像素在原图像中的边缘方向包括：
    分别根据公式(2)和(3)计算原图像中待插值像素邻域内若干像素的水平梯度gH(i,j)和垂直梯度gV(i,j)：
    gH(i,j)＝IL(i-1,j+1)+2*IL(i,j+1)+IL(i+1,j+1)  (2)
    -IL(i-1,j-1)-2*IL(i,j-1)-IL(i+1,j-1)
    gV(i,j)＝IL(i-1,j-1)+2*IL(i-1,j)+IL(i-1,j+1)  (3)
    -IL(i+1,j-1)-2*IL(i+1,j)-IL(i+1,j+1)
    分别根据所述待插值像素在原图像中的位置与所述邻域内像素的水平梯度和垂直梯度利用双线性插值也即根据公式(4)和(5)确定待插值像素的水平梯度gH(iL,jL)和垂直梯度gV(iL,jL)，则待插值像素的边缘方向即是 所述待插值像素的梯度的方向的垂直方向(gV(iL,jL),-gH(iL,jL))：
    gH(iL,jL)＝(1-dx)*(1-dy)*gH(i,j)+dx*(1-dy)*gH(i,j+1)  (4)
    +(1-dx)*dy*gH(i+1,j)+dx*dy*gH(i+1,j+1)
    gV(iL,jL)＝(1-dx)*(1-dy)*gV(i,j)+dx*(1-dy)*gV(i,j+1)  (5)
    +(1-dx)*dy*gV(i+1,j)+dx*dy*gV(i+1,j+1)
    其中，IL(i-1,j+1)、IL(i,j+1)、IL(i+1,j+1)、IL(i-1,j-1)、IL(i,j-1)、IL(i+1,j-1)、IL(i-1,j)、IL(i+1,j)分别表示在原图像中待插值像素邻域内八个像素的像素值。
  6. 根据权利要求1所述的基于边缘检测的图像插值方法，其特征在于，所述确定待插值像素在原图像中的边缘方向包括：
    选取待插值像素邻域内任意大小的窗口，确定窗口内全部像素的水平梯度gH(i,j)和垂直梯度gV(i,j)，从而确定待插值像素邻域内窗口中全部像素的协方差矩阵M：
    Figure PCTCN2015075709-appb-100009
    计算所述协方差矩阵的特征值和特征向量，则确定较小特征值对应的特征向量v为所述边缘方向，也即：
    Figure PCTCN2015075709-appb-100010
    其中，
    Figure PCTCN2015075709-appb-100011
    表示所述协方差矩阵的较小特征值对应的特征向量。vx表示边缘方向的水平分量，vy表示边缘方向的垂直分量。
  7. 据权利要求6所述的基于边缘检测的图像插值方法，其特征在于，所述确定待插值像素在原图像中的边缘方向还包括：
    根据公式(6)改进所述协方差矩阵获得改进后的协方差矩阵M'：
    Figure PCTCN2015075709-appb-100012
    其中，w(i,j)的取值采用双线性插值的取法，即为：
    w(i-1,j-2)＝(1-dx)*(1-dy)，w(i-1,j-1)＝(1-dy)，w(i-1,j)＝(1-dy)，
    w(i-1,j+1)＝(1-dy)，w(i-1,j+2)＝(1-dy)，w(i-1,j+3)＝dx*(1-dy)；
    w(i,j-2)＝(1-dx)，w(i,j-1)＝1，w(i,j)＝1，w(i,j+1)＝1，w(i,j+2)＝1，
    w(i,j+3)＝dx；w(i+1,j-2)＝(1-dx)，w(i+1,j-1)＝1，w(i+1,j)＝1，w(i+1,j+1)＝1，
    w(i+1,j+2)＝1，w(i+1,j+3)＝dx；w(i+2,j-2)＝(1-dx)*dy，w(i+2,j-1)＝dy，
    w(i+2,j)＝dy，w(i+2,j+1)＝dy，w(i+2,j+2)＝dy，w(i+2,j+3)＝dx*dy。
  8. 一种基于边缘检测的图像插值系统，其特征在于，包括：
    坐标计算单元，用于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；
    方向计算单元，用于确定待插值像素在原图像中的边缘方向；
    交点计算单元，用于在边缘方向的斜率不小于第一阈值时，计算原图像中待插值像素邻域内若干行和/或若干列被待插值像素和边缘方向确定的直线所截的若干行交点和/或若干列交点的位置；
    边缘插值滤波单元，用于根据行交点法和/或列交点法进行插值，具体用于利用一维插值法根据原图像中所述行交点和/或列交点邻域内像素的值确定所述行交点和/或列交点的像素值、和对所确定的待插值像素邻域内的行交点和/或列交点的像素值进行一维滤波，获得待插值像素的值，并对原图像进行插值。
  9. 根据权利要求8所述的基于边缘检测的图像插值系统，其特征在于，还包括：
    非边缘插值单元，用于在边缘方向的斜率的绝对值小于设定阈值时，根据非边缘插值法进行插值；
    融合单元，用于对所述行交点法和/或列交点法插值得到的结果与所述非边缘插值法得到的结果进行融合从而获得插值后的图像。
  10. 根据权利要求8所述的基于边缘检测的图像插值系统，其特征在于，所述方向计算单元具体用于：分别根据公式(2)和(3)计算原图像中待插值像素邻域内若干像素的水平梯度gH(i,j)和垂直梯度gV(i,j)：
    gH(i,j)＝IL(i-1,j+1)+2*IL(i,j+1)+IL(i+1,j+1)  (2)
    -IL(i-1,j-1)-2*IL(i,j-1)-IL(i+1,j-1)
    gV(i,j)＝IL(i-1,j-1)+2*IL(i-1,j)+IL(i-1,j+1)  (3)
    -IL(i+1,j-1)-2*IL(i+1,j)-IL(i+1,j+1)
    和分别根据所述待插值像素在原图像中的位置与所述邻域内像素的水平梯度和垂直梯度利用双线性插值也即根据公式(4)和(5)确定待插值像素的水平梯度gH(iL,jL)和垂直梯度gV(iL,jL)，则待插值像素的边缘方向即是所述待插值像素的梯度的方向的垂直方向(gV(iL,jL),-gH(iL,jL))：
    gH(iL,jL)＝(1-dx)*(1-dy)*gH(i,j)+dx*(1-dy)*gH(i,j+1)  (4)
    +(1-dx)*dy*gH(i+1,j)+dx*dy*gH(i+1,j+1)
    gV(iL,jL)＝(1-dx)*(1-dy)*gV(i,j)+dx*(1-dy)*gV(i,j+1)  (5)
    +(1-dx)*dy*gV(i+1,j)+dx*dy*gV(i+1,j+1)
    其中，IL(i-1,j+1)、IL(i,j+1)、IL(i+1,j+1)、IL(i-1,j-1)、IL(i,j-1)、IL(i+1,j-1)、IL(i-1,j)、IL(i+1,j)分别表示在原图像中待插值像素邻域内八个像素的像素值。
PCT/CN2015/075709 2015-04-01 2015-04-01 基于边缘检测的图像插值方法及系统 WO2016154970A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2015/075709 WO2016154970A1 (zh) 2015-04-01 2015-04-01 基于边缘检测的图像插值方法及系统

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2015/075709 WO2016154970A1 (zh) 2015-04-01 2015-04-01 基于边缘检测的图像插值方法及系统

Publications (1)

Publication Number Publication Date
WO2016154970A1 true WO2016154970A1 (zh) 2016-10-06

Family

ID=57003804

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2015/075709 WO2016154970A1 (zh) 2015-04-01 2015-04-01 基于边缘检测的图像插值方法及系统

Country Status (1)

Country Link
WO (1) WO2016154970A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114494306A (zh) * 2022-01-27 2022-05-13 辽宁师范大学 边缘梯度协方差引导的甲骨卜辞文字轮廓修复方法

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050073607A1 (en) * 2003-10-02 2005-04-07 Samsung Electronics Co., Ltd. Image adaptive deinterlacing method and device based on edge
US20050249417A1 (en) * 2004-05-06 2005-11-10 Dong-Seob Song Edge detecting method
CN101790069A (zh) * 2010-03-09 2010-07-28 周艇 一种基于图像边缘方向的尺度变换方法
CN101931821A (zh) * 2010-07-21 2010-12-29 中兴通讯股份有限公司 一种视频传输的差错控制方法及系统
CN102045530A (zh) * 2010-12-30 2011-05-04 北京中科大洋科技发展股份有限公司 一种基于边缘检测的运动自适应的去交织方法
CN103366342A (zh) * 2013-07-02 2013-10-23 天津大学 应用于视频图像放大的分段线性插值方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050073607A1 (en) * 2003-10-02 2005-04-07 Samsung Electronics Co., Ltd. Image adaptive deinterlacing method and device based on edge
US20050249417A1 (en) * 2004-05-06 2005-11-10 Dong-Seob Song Edge detecting method
CN101790069A (zh) * 2010-03-09 2010-07-28 周艇 一种基于图像边缘方向的尺度变换方法
CN101931821A (zh) * 2010-07-21 2010-12-29 中兴通讯股份有限公司 一种视频传输的差错控制方法及系统
CN102045530A (zh) * 2010-12-30 2011-05-04 北京中科大洋科技发展股份有限公司 一种基于边缘检测的运动自适应的去交织方法
CN103366342A (zh) * 2013-07-02 2013-10-23 天津大学 应用于视频图像放大的分段线性插值方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
LAI, YUESHEN ET AL.: "The Optimization of Sub-Pixel Edge Detection Algorithm with Gauss Interpolation", JOURNAL OF XI'AN TECHNOLOGICAL UNIVERSITY, vol. 32, no. 10, 31 October 2012 (2012-10-31) *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114494306A (zh) * 2022-01-27 2022-05-13 辽宁师范大学 边缘梯度协方差引导的甲骨卜辞文字轮廓修复方法
CN114494306B (zh) * 2022-01-27 2024-04-05 辽宁师范大学 边缘梯度协方差引导的甲骨卜辞文字轮廓修复方法

Similar Documents

Publication Publication Date Title
CN106651938B (zh) 一种融合高分辨率彩色图像的深度图增强方法
CN103034973B (zh) 基于双三次插值的自适应图像缩放方法
JP6142611B2 (ja) ステレオマッチングのための方法及びステレオマッチングのためのシステム
JP6016061B2 (ja) 画像生成装置、画像表示装置及び画像生成方法並びに画像生成プログラム
TWI543586B (zh) 影像增強的方法及其影像處理裝置與電腦程式產品
US8494308B2 (en) Image upscaling based upon directional interpolation
CN104700360B (zh) 基于边缘自适应的图像缩放方法及系统
US20110188749A1 (en) Image processing method for boundary resolution enhancement
KR20160064953A (ko) 영상 처리 장치, 영상 처리 방법, 및 프로그램
US20190096031A1 (en) Image interpolation methods and related image interpolation devices thereof
CN104700361B (zh) 基于边缘检测的图像插值方法及系统
JP2002525723A (ja) ディジタル画像をズームする方法及び装置
Jeong et al. Multi-frame example-based super-resolution using locally directional self-similarity
US8830395B2 (en) Systems and methods for adaptive scaling of digital images
CN104537627B (zh) 一种深度图像的后处理方法
US10410326B2 (en) Image anti-aliasing system
CN102750668A (zh) 结合局部方向特征的数字图像三倍插值放大方法
CN102682424B (zh) 基于边缘方向差值的图像放大处理方法
CN110349090B (zh) 一种基于牛顿二阶插值的图像缩放方法
CN109325909B (zh) 一种图像放大方法和图像放大装置
WO2016154970A1 (zh) 基于边缘检测的图像插值方法及系统
JP4868249B2 (ja) 映像信号処理装置
Kang et al. Real-time super-resolution for digital zooming using finite kernel-based edge orientation estimation and truncated image restoration
Ousguine et al. A new image interpolation using gradient-orientation and cubic spline interpolation
WO2016154966A1 (zh) 基于边缘自适应的图像缩放方法及系统

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: 15886940

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: 15886940

Country of ref document: EP

Kind code of ref document: A1