WO2016029555A1 - 图像插值方法和装置 - Google Patents
图像插值方法和装置 Download PDFInfo
- Publication number
- WO2016029555A1 WO2016029555A1 PCT/CN2014/091076 CN2014091076W WO2016029555A1 WO 2016029555 A1 WO2016029555 A1 WO 2016029555A1 CN 2014091076 W CN2014091076 W CN 2014091076W WO 2016029555 A1 WO2016029555 A1 WO 2016029555A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- pixel
- interpolated
- edge
- image
- interpolation
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000004364 calculation method Methods 0.000 claims abstract description 22
- 229910003460 diamond Inorganic materials 0.000 description 3
- 239000010432 diamond Substances 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
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 present disclosure relates to the field of image interpolation techniques, and in particular, to an image interpolation method and apparatus.
- the existing image interpolation algorithm only considers the relationship between the position of the pixel to be interpolated and the surrounding control pixel, and does not consider the characteristics of the image itself.
- the high frequency part of the image that is, the edge part, is easily lost in the traditional weighting operation, but the human eye It is sensitive to the edge of the image, and the overall effect of interpolation is susceptible to edge blur. How to better preserve the high-frequency region of the image without affecting the low-frequency region of the image is the key to solve the problem.
- the main purpose of the present disclosure is to provide an image interpolation method and apparatus to solve the problem of image edge blur caused by unrelated pixel points participating in weighting when the image interpolation algorithm interpolates in the image edge region.
- the present disclosure provides an image interpolation method, the image including an edge region, a texture region, and a flat region; the image interpolation method includes:
- M original pixel points adjacent to the pixel to be interpolated are selected along the edge direction, and the M original pixel points are used as control points for interpolation calculation, and M is an integer greater than 1.
- the step of determining an edge direction of the pixel to be interpolated specifically includes: constructing an edge energy function E edge with respect to ⁇ according to N original pixel points adjacent to the pixel to be interpolated selected according to a predetermined rule ( ⁇ ), wherein ⁇ is an angle corresponding to an edge direction, and N is an integer greater than 1;
- the edge direction of the pixel to be interpolated is determined according to the edge energy function E edge ( ⁇ ).
- edge energy function E edge ( ⁇ ) is as follows:
- ⁇ is the angle between the edge direction and the horizontal direction
- i, j are the rows and columns of the original pixel points closest to the upper left corner of the pixel to be interpolated;
- d is the value corresponding to the edge direction of the pixel to be interpolated; d is 1 when ⁇ is 45°, d is -1 when ⁇ is 135°, d is 0 when ⁇ is 90°, and 0 is 30 when ⁇ is 90° When d is 2, when ⁇ is 150°, d is -2;
- v is the horizontal pixel position interval in the edge direction
- h is the vertical pixel position interval in the edge direction
- I(i+d+v, j+d+h) is the gray value of the j+d+h column of the i+d+v row;
- I(i+v, j+h) is the gray value of the j+hth column pixel of the i+vth line.
- determining the edge direction of the pixel to be interpolated according to the edge energy function E edge ( ⁇ ) specifically includes:
- the ⁇ corresponding to the minimum edge energy function E edge ( ⁇ ) is the angle between the edge direction of the pixel to be interpolated and the horizontal direction to be determined.
- the range of parameters for calculating the edge energy function E edge ( ⁇ ) is
- the N original pixel points that are selected according to the predetermined rule and are adjacent to the pixel to be interpolated include:
- interpolation is performed by using a bicubic interpolation method.
- the selecting the M original pixel points adjacent to the pixel to be interpolated along the edge direction, and performing the interpolation calculation by using the M original pixel points as the control points includes:
- the image interpolation method of the present disclosure further includes:
- the noise of the pixel to be interpolated is detected, and the corresponding filter is selected according to the noise for interpolation.
- the image interpolation method of the present disclosure further includes:
- the interpolation calculation is performed by the cubic convolution interpolation method.
- the present disclosure also provides an image interpolation apparatus, the image including an edge area, a texture area, and a flat area; the image interpolation apparatus includes:
- a judging module configured to perform regional judgment on the pixels to be interpolated
- an interpolation module configured to: when the determining module determines that the pixel to be interpolated is in an edge region, determine an edge direction of the pixel to be interpolated, and select M originals adjacent to the pixel to be interpolated along the edge direction The pixel is interpolated with the M original pixel points as control points, and M is an integer greater than 1.
- the determining module is further configured to determine, when the image is interpolated, whether the pixel to be interpolated is in a texture area;
- the interpolation module is further configured to: when the determining module determines that the pixel to be interpolated is in a texture area In the domain, the noise of the pixel to be interpolated is detected, and the corresponding filter is selected according to the noise for interpolation.
- the determining module is further configured to determine, when the image is interpolated, whether the pixel to be interpolated is in a flat region;
- the interpolation module is further configured to perform interpolation calculation by using a cubic convolution interpolation method when the determining module determines that the pixel to be interpolated is in a flat region.
- the image interpolation method and apparatus can solve the image edge blur problem caused by the unrelated pixel points participating in the weighting of the image edge region interpolation based on the edge alienation interpolation; and the present disclosure is to press the edge region by the image.
- the texture area and the flat area are divided. Different areas are interpolated to improve the sharpness and realism of the interpolated image. While retaining the high frequency information of the image, the effect of the low frequency area of the image is not affected.
- FIG. 1 is a flowchart of an image interpolation method according to an embodiment of the present disclosure
- Figure 2 is a schematic view showing the division of the edge direction of the human eye
- FIG. 3 is a schematic diagram of selecting control points along an edge direction in an image interpolation method according to an embodiment of the present disclosure
- FIG. 4 is a structural block diagram of an image interpolation apparatus according to an embodiment of the present disclosure.
- the present disclosure provides an image interpolation method including an edge region, a texture region, and a flat region. As shown in FIG. 1, the image interpolation method according to the embodiment of the present disclosure includes:
- Step 11 Perform area determination on the interpolated pixel points
- the image gradation change frequency of the edge region is higher than the non-edge region, it can be determined, for example, by filtering whether the pixel to be interpolated is in the edge region.
- the non-edge area may include a texture area and a flat area.
- the edge region can be understood as a region where the edge contour gray frequency changes relatively.
- the texture region can be understood as a region where the edge contour gray frequency changes little but changes frequently, and the flat region can be understood as a region where the gray frequency changes little. .
- Step 22 If the pixel to be interpolated is in the edge region, then
- M original pixel points adjacent to the pixel to be interpolated are selected along the edge direction, and the M original pixel points are used as control points for interpolation calculation, and M is an integer greater than 1.
- the image interpolation method can solve the image edge blur problem caused by the unrelated pixel points participating in the weighting of the image interpolation by the traditional image interpolation algorithm based on the edge alienation interpolation.
- the step of determining an edge direction of the pixel to be interpolated includes:
- N is an integer greater than 1
- constructing an edge energy function E edge ( ⁇ ) related to the angle ⁇ corresponding to the edge direction to be determined is as follows:
- ⁇ is the angle between the edge direction and the horizontal direction
- i, j are the rows and columns of the original pixel points closest to the upper left corner of the pixel to be interpolated;
- d is the value corresponding to the edge direction of the pixel to be interpolated; wherein d is 1 when ⁇ is 45°, d is ⁇ 1 when ⁇ is 135°, and d is 0 when ⁇ is 90°, when ⁇ is When it is 30°, d is 2, and when ⁇ is 150°, d is -2;
- v is the horizontal pixel position interval in the edge direction
- h is the vertical pixel position interval in the edge direction
- I(i+d+v, j+d+h) is the gray value of the j+d+h column of the i+d+v row;
- I(i+v, j+h) is the gray value of the j+hth column pixel of the i+vth line.
- the calculated E edge ( ⁇ ) in each direction is compared with the minimum edge energy.
- the ⁇ corresponding to the function E edge ( ⁇ ) is the angle between the edge direction of the pixel to be interpolated and the horizontal direction to be determined.
- C min(E edge ( ⁇ )
- C is the correlation of the pixel to be interpolated in its edge direction d. Since the value of the value d corresponding to the edge direction of the pixel to be interpolated is discrete, it can be understood that the edge direction calculated according to the above method and the actual edge direction are erroneous. The above error can be reflected by the correlation C. It should be understood by those skilled in the art that the smaller the value of the correlation C is, the closer the calculated edge direction is to the actual edge direction.
- edge direction A detailed description of the edge direction is as follows.
- a plurality of circles represent pixel points.
- the rest of the more oblique direction detection is very difficult, and there are few appearances in the image, and the impact on the overall quality of the image is minimal. Therefore, in the embodiment of the present invention, only the direction-2, the direction-1, the direction 0, the direction 1 and the direction 2 are calculated, so that the calculation amount can be reduced as much as possible on the basis of ensuring the quality of the interpolated image.
- Improve the speed of computing The direction 0 can be regarded as a flat area, and the direction-2, the direction-1, the direction 1 and the direction 2 all represent an edge area, and the pixel to be inserted in the edge area needs to be interpolated along the edge direction.
- the E edge ( ⁇ ) it is also possible to calculate the E edge ( ⁇ ) by using three consecutively adjacent three pixel points respectively located on the upper and lower sides of the pixel to be interpolated, that is, a total of six pixels, which reduces the calculation amount. At the same time, noise interference can be eliminated to some extent, and the detected edge direction accuracy is ensured.
- the setting of the edge energy function is not limited to the example provided in the above embodiment, and any edge energy function by which the pixel to be interpolated can be detected can be applied.
- f0 indicates the pixel to be interpolated
- f1, f2, f3, and f4 are the pixels in the original pixel row of the row on f0
- f5, f6, f7, and f8 are the original pixels of the next row in f0.
- edge direction of f0 is the direction 0, f2, f3, f6, and f7 selected in the direction 0 (i.e., adjacent 4 original pixel points in the edge direction) are interpolated as control points.
- the rays from f6 to f7 are taken as the x-axis, and the rays from f6 to f2 are taken as the y-axis (not shown) to form a Cartesian coordinate system. Subsequent calculations are similar to those in the diamond coordinate system and are omitted here.
- the image interpolation method of the embodiment of the present disclosure further includes:
- the noise of the pixel to be interpolated is detected, and the corresponding filter is selected according to the noise for interpolation.
- the image interpolation method of the embodiment of the present disclosure further includes:
- the interpolation calculation is performed by the cubic convolution interpolation method.
- the image interpolation method and apparatus divides an image by an edge region, a texture region, and a flat region, and different regions adopt different interpolation methods to improve the sharpness and realism of the image after interpolation, and preserve the image.
- the high frequency information ensures that the effect of the low frequency region of the image is not affected.
- the present disclosure also provides an image interpolation apparatus, the image including an edge area, a texture area, and a flat area; as shown in FIG. 4, the image interpolation apparatus includes:
- the determining module 41 is configured to perform area determination on the pixel to be interpolated
- an interpolation module 42 configured to: when the determining module 41 determines that the pixel to be interpolated is in an edge region, determine an edge direction of the pixel to be interpolated, and select an M adjacent to the pixel to be interpolated along the edge direction
- the original pixel points are interpolated with the M original pixel points as control points, and M is an integer greater than 1.
- the determining module is further configured to determine, when the image is interpolated, whether the pixel to be interpolated is in a texture area;
- the interpolation module is further configured to: when the determining module determines that the pixel to be interpolated is in a texture region, detect noise of the pixel to be interpolated, and select a corresponding filter according to the noise to perform interpolation.
- the determining module is further configured to determine, when the image is interpolated, whether the pixel to be interpolated is in a flat region;
- the interpolation module is further configured to perform interpolation calculation by using a cubic convolution interpolation method when the determining module determines that the pixel to be interpolated is in a flat region.
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
本公开提供了一种图像插值方法和装置。所述图像插值方法用于对图像进行插值，所述图像包括边缘区域、纹理区域和平坦区域；所述图像插值方法包括：对待插值像素点进行区域判断；若待插值像素点处于边缘区域，确定该待插值像素点的边缘方向，沿该边缘方向选取与该待插值像素点相邻的M个原始像素点，以该M个原始像素点作为控制点进行插值计算，M为大于1的整数。
Description
相关申请的交叉引用
本申请主张在2014年8月25日在中国提交的中国专利申请号No.201410422383.1的优先权，其全部内容通过引用包含于此。
本公开涉及图像插值技术领域，尤其涉及一种图像插值方法和装置。
现有的图像插值算法只考虑待插值像素点位置与周围控制像素点的关系，并没有考虑图像本身的特征，图像的高频地方即边缘部分，在传统的加权运算中容易损失，但人眼对图像边缘比较敏感，插值的整体效果易受边缘模糊影响，如何较好的保留图像高频区域，又不影响图像低频区域效果是解决该问题的关键。
发明内容
本公开的主要目的在于提供一种图像插值方法和装置，以解决传统图像插值算法在图像边缘区域插值时不相关像素点参与加权导致的图像边缘模糊问题。
为了达到上述目的，本公开提供了一种图像插值方法，所述图像包括边缘区域、纹理区域和平坦区域；所述图像插值方法包括：
对待插值像素点进行区域判断；
若待插值像素点处于边缘区域，则
确定该待插值像素点的边缘方向；
沿该边缘方向选取与该待插值像素点相邻的M个原始像素点，以该M个原始像素点作为控制点进行插值计算，M为大于1的整数。
可选地，所述确定该待插值像素点的边缘方向的步骤具体包括：根据按照预定规则选取的与该待插值像素点相邻的N个原始像素点，构造关于α的
边缘能量函数Eedge(α)，其中，α为与边缘方向对应的角度，N为大于1的整数；
根据该边缘能量函数Eedge(α)确定所述待插值像素点的边缘方向。
可选地，所述边缘能量函数Eedge(α)如下：
其中，
α为边缘方向与水平方向的夹角；
i，j分别为待插值像素点左上角最邻近的原始像素点所在的行和列；
d为该待插值像素点的边缘方向所对应的数值；当α为45°时d为1，当α为135°时d为-1，当α为90°时d为0，当α为30°时d为2，当α为150°时d为-2；
v为在边缘方向上的水平像素位置间隔；
h为在边缘方向上的垂直像素位置间隔；
I(i+d+v，j+d+h)为第i+d+v行第j+d+h列像素的灰度值；
I(i+v，j+h)为第i+v行第j+h列像素的灰度值。
可选地，所述根据该边缘能量函数Eedge(α)确定所述待插值像素点的边缘方向具体包括：
对各个方向上计算得到的Eedge(α)进行比较，与最小的边缘能量函数Eedge(α)相对应的α即为待确定的该待插值像素点的边缘方向与水平方向的夹角。
可选地，计算所述边缘能量函数Eedge(α)的参数的范围是
d∈{-2，-1，0，1，2}，v∈{-4，-2，0，2，4}，h∈{-2，-1，0，1，2}。
可选地，所述按照预定规则选取的与该待插值像素点相邻的N个原始像素点，具体包括：
分别位于待插值像素点上下两侧的各自连续相邻的三个像素点，共六个
像素点。
可选地，，当确定的所述待插值像素点的边缘方向与水平方向的夹角α为0°时，利用双三次插值方式进行插值。
可选地，所述沿该边缘方向选取与该待插值像素点相邻的M个原始像素点，以该M个原始像素点作为控制点进行插值计算具体包括：
在该待插值像素点上、下两个相邻原始像素行，沿该边缘方向选取最邻近的4个原始像素点，该待插值像素点的灰度值由该4个原始像素点作为控制点进行双线性插值计算获得。
可选地，本公开所述的图像插值方法还包括：
在对所述图像进行插值时，判断待插值像素点是否处于纹理区域；
当判断到待插值像素点处于纹理区域时，检测该待插值像素点的噪声，根据该噪声选取对应的滤波器进行插值。
可选地，本公开所述的图像插值方法还包括：
在对所述图像进行插值时，判断待插值像素点是否处于平坦区域；
当判断到待插值像素点处于平坦区域时，采用三次卷积插值法进行插值计算。
本公开还提供了一种图像插值装置，所述图像包括边缘区域、纹理区域和平坦区域；所述图像插值装置包括：
判断模块，用于对待插值像素点进行区域判断；
以及，插值模块，用于当所述判断模块判断到待插值像素点处于边缘区域时，确定该待插值像素点的边缘方向，沿该边缘方向选取与该待插值像素点相邻的M个原始像素点，以该M个原始像素点作为控制点进行插值计算，M为大于1的整数。
可选地，所述判断模块，还用于在对所述图像进行插值时，判断待插值像素点是否处于纹理区域；
所述插值模块，还用于当所述判断模块判断到待插值像素点处于纹理区
域时，检测该待插值像素点的噪声，根据该噪声选取对应的滤波器进行插值。
可选地，所述判断模块，还用于在对所述图像进行插值时，判断待插值像素点是否处于平坦区域；
所述插值模块，还用于当所述判断模块判断到待插值像素点处于平坦区域时，采用三次卷积插值法进行插值计算。
本公开所述的图像插值方法和装置，基于边缘异化插值，可以解决传统图像插值算法在图像边缘区域插值时不相关像素点参与加权导致的图像边缘模糊问题；并且本公开通过将图像按边缘区域、纹理区域和平坦区域进行划分，不同的区域采用不同插值方法，以提高插值后图像的清晰度和真实度，在保留图像高频信息的同时又能保证图像低频区域的效果不受影响。
图1是本公开实施例所述的图像插值方法的流程图；
图2是人眼敏感的边缘方向的划分示意图；
图3是本公开实施例所述的图像插值方法中沿边缘方向选取控制点的示意图；
图4是本公开实施例所述的图像插值装置的结构框图。
本公开提供了一种图像插值方法，所述图像包括边缘区域、纹理区域和平坦区域。如图1所示，本公开实施例所述的图像插值方法包括：
步骤11：对待插值像素点进行区域判断；
具体地，因为边缘区域的图像灰度变化频率高于非边缘区域，所以例如可以通过滤波来判断待插值像素点是否处于边缘区域。
这里，非边缘区可以包括纹理区域和平坦区域。边缘区域可以理解为边缘轮廓灰度频率变化比较大的区域，纹理区域可以理解为边缘轮廓灰度频率变化较小的但变化频繁的区域，而平坦区域可以理解为灰度频率变化很小的区域。对于待插值像素点处于哪一个区域，具体的判断分类方法有很多种。
例如，可以用Cannny，Sobel，Laplace等方法判断待插值像素点处于哪一个区域，其中，当将边缘提取阈值参数设置为较小的值(例如threshlod＝3)时，可以得到边缘区域和纹理区域；当将边缘提取阈值参数设置为较大的值(例如threshlod＝5)时，得到的仅为边缘区域。本领域人员应当理解的是，不同的边缘提取算法，阈值参数的设置也不相同。
步骤22：若待插值像素点处于边缘区域，则
确定该待插值像素点的边缘方向；
沿该边缘方向选取与该待插值像素点相邻的M个原始像素点，以该M个原始像素点作为控制点进行插值计算，M为大于1的整数。
本公开实施例所述的图像插值方法，基于边缘异化插值，可以解决传统图像插值算法在图像边缘区域插值时不相关像素点参与加权导致的图像边缘模糊问题。
在一个实施例中，所述确定该待插值像素点的边缘方向的步骤包括：
选取与该待插值像素点相邻的N个原始像素点，N为大于1的整数，构造与待定边缘方向对应角度α有关的边缘能量函数Eedge(α)如下：
其中，
α为边缘方向与水平方向的夹角；
i，j分别为待插值像素点左上角最邻近的原始像素点所在的行和列；
d为该待插值像素点的边缘方向所对应的数值；其中，当α为45°时d为1，当α为135°时d为-1，当α为90°时d为0，当α为30°时d为2，当α为150°时d为-2；
v为在边缘方向上的水平像素位置间隔；
h为在边缘方向上的垂直像素位置间隔；
I(i+d+v，j+d+h)为第i+d+v行第j+d+h列像素的灰度值；
I(i+v，j+h)为第i+v行第j+h列像素的灰度值。
根据图像中沿着边沿方向的像素值变化率很小，而垂直于边沿方向的像素值变化率很快的特性，对各个方向上计算得到的Eedge(α)进行比较，与最小的边缘能量函数Eedge(α)相对应的α即为待确定的该待插值像素点的边缘方向与水平方向的夹角。
令：C＝min(Eedge(α)|α∈{30，45，90，135，150})，C是所述待插值像素点在其边缘方向d上的相关度。由于作为待插值像素点的边缘方向所对应的数值d的取值是离散的，因此，可以理解的是，根据上述方法计算出的边缘方向和实际的边缘方向是有误差的。而上述误差可以通过相关度C来体现，本领域人员应当理解的是，相关度C的值越小，则计算出的边缘方向和实际的边缘方向越接近。
关于边缘方向的具体说明如下。在图2中，多个圆圈表示的是像素点。其余的更加倾斜的方向检测的难度很大，而且在图像中出现的很少，对图像整体质量的影响微乎其微。因此，可选地，本发明实施例中只对方向-2、方向-1、方向0、方向1和方向2进行计算，这样可以在保证插值图像质量的基础上尽可能地减小计算量，提高运算速度。其中，方向0可以看作平坦区域，方向-2、方向-1、方向1和方向2均表示边缘区域，在边缘区域内待插入的像素点需要沿着边缘方向进行插值计算。
因此，可选地，d∈{-2，-1，0，1，2}，v∈{-4，-2，0，2，4}，h∈{-2，-1，0，1，2}。
在一个实施例中，还可以用分别位于待插值像素点上下两侧的各自连续相邻的三个像素点，即共六个像素点，进行Eedge(α)的计算，在降低了计算量的同时，还可以在一定程度上消除噪声的干扰，并确保检测出的边缘方向准确性。
在实际操作时，边缘能量函数的设定不限于上述实施例中提供的示例，任何可以通过其检测出待插值像素点的边缘能量函数都可以适用。
所述沿该边缘方向选取与该待插值像素点相邻的M个原始像素点，以该
M个原始像素点作为控制点进行插值计算的步骤具体包括：
在该待插值像素点上、下两个相邻原始像素行，沿该边缘方向选取最邻近的4个原始像素点，该待插值像素点的灰度值由该4个原始像素点作为控制点进行双线性插值计算获得。
参考图3，以M＝4为例的具体描述如下。
在图3中，f0标示的是待插值像素点，f1、f2、f3和f4是在f0上一行的原始像素行中的像素点，f5、f6、f7和f8是在f0下一行的原始像素行中的像素点。
如图3所示，当f0的边缘方向为方向1时，沿方向1选取f3、f4、f6和f7，以f6为原点，从f6向f7的射线作为x’轴，从f6向f3的射线作为y’轴，形成菱形坐标系，
以f0为顶点并与y’轴平行的线与x’轴的交点与f6的距离为Δx；
以f0为顶点并与x’轴平行的线与y’轴的交点与f6的距离为Δy；
则f0在该菱形坐标系中的坐标为(Δx，Δy)；
将(Δx，Δy)，像素点f3、f4、f6和f7(即该边缘方向上的相邻的4个原始像素点)的坐标及其灰度值代入双线性插值计算单元进行边缘区域的插值计算，以计算f0的灰度。
在图3中，当f0的边缘方向为方向0时，沿方向0选取的f2、f3、f6和f7(即该边缘方向上的相邻的4个原始像素点)作为控制点进行插值计算。在这种情况下，从f6向f7的射线作为x轴，从f6向f2的射线作为y轴(图中未示出)，形成直角坐标系。后续的计算方法与在菱形坐标系中类似，此处省去。
可选地，本公开实施例所述的图像插值方法，还包括：
在对所述图像进行插值时，判断待插值像素点是否处于纹理区域；
当判断到待插值像素点处于纹理区域时，检测该待插值像素点的噪声，根据该噪声选取对应的滤波器进行插值。
可选地，本公开实施例所述的图像插值方法，还包括：
在对所述图像进行插值时，判断待插值像素点是否处于平坦区域；
当判断到待插值像素点处于平坦区域时，采用三次卷积插值法进行插值计算。
在实际操作时，只需对处于平坦区域的待插值像素点采用常规的插值方法进行插值计算即可，并不限于采用三次卷积插值法。
本公开实施例所述的图像插值方法和装置通过将图像按边缘区域、纹理区域和平坦区域进行划分，不同的区域采用不同插值方法，以提高插值后图像的清晰度和真实度，在保留图像高频信息的同时又能保证图像低频区域的效果不受影响。
本公开还提供了一种图像插值装置，所述图像包括边缘区域、纹理区域和平坦区域；如图4所示，所述图像插值装置包括：
判断模块41，用于对待插值像素点进行区域判断；
以及，插值模块42，用于当所述判断模块41判断到待插值像素点处于边缘区域时，确定该待插值像素点的边缘方向，沿该边缘方向选取与该待插值像素点相邻的M个原始像素点，以该M个原始像素点作为控制点进行插值计算，M为大于1的整数。
可选地，所述判断模块，还用于在对所述图像进行插值时，判断待插值像素点是否处于纹理区域；
所述插值模块，还用于当所述判断模块判断到待插值像素点处于纹理区域时，检测该待插值像素点的噪声，根据该噪声选取对应的滤波器进行插值。
可选地，所述判断模块，还用于在对所述图像进行插值时，判断待插值像素点是否处于平坦区域；
所述插值模块，还用于当所述判断模块判断到待插值像素点处于平坦区域时，采用三次卷积插值法进行插值计算。
以上所述是本公开的优选实施方式，应当指出，对于本技术领域的普通
技术人员来说，在不脱离本公开所述原理的前提下，还可以作出若干改进和润饰，这些改进和润饰也应视为本公开的保护范围。
Claims (13)
- 一种图像插值方法，所述图像包括边缘区域、纹理区域和平坦区域；其中，所述图像插值方法包括：对待插值像素点进行区域判断；若待插值像素点处于边缘区域，则确定该待插值像素点的边缘方向；沿该边缘方向选取与该待插值像素点相邻的M个原始像素点，以该M个原始像素点作为控制点进行插值计算，M为大于1的整数。
- 如权利要求1所述的图像插值方法，其中，所述确定该待插值像素点的边缘方向的步骤具体包括：根据按照预定规则选取的与该待插值像素点相邻的N个原始像素点，构造关于α的边缘能量函数Eedge(α)，其中，α为与边缘方向对应的角度，N为大于1的整数；根据该边缘能量函数Eedge(α)确定所述待插值像素点的边缘方向。
- 如权利要求2所述的图像插值方法，其中，所述边缘能量函数Eedge(α)如下：
其中，α为边缘方向与水平方向的夹角；i，j分别为待插值像素点左上角最邻近的原始像素点所在的行和列；d为该待插值像素点的边缘方向所对应的数值；当α为45°时d为1，当α为135°时d为-1，当α为90°时d为0，当α为30°时d为2，当α为150°时d为-2；v为在边缘方向上的水平像素位置间隔；h为在边缘方向上的垂直像素位置间隔；I(i+d+v，j+d+h)为第i+d+v行第j+d+h列像素的灰度值；I(i+v，j+h)为第i+v行第j+h列像素的灰度值。 - 如权利要求3所述的图像插值方法，其中，所述根据该边缘能量函数Eedge(α)确定所述待插值像素点的边缘方向的步骤具体包括：对各个方向上计算得到的Eedge(α)进行比较，与最小的边缘能量函数Eedge(α)相对应的α即为待确定的该待插值像素点的边缘方向与水平方向的夹角。
- 如权利要求3所述的图像插值方法，其中，计算所述边缘能量函数Eedge(α)的参数的范围是d∈{-2，-1，0，1，2}，v∈{-4，-2，0，2，4}，h∈{-2，-1，0，1，2}。
- 如权利要求2所述的图像插值方法，其中，所述按照预定规则选取的与该待插值像素点相邻的N个原始像素点，具体包括：分别位于待插值像素点上下两侧的各自连续相邻的三个像素点，共六个像素点。
- 如权利要求3所述的图像插值方法，其中，当确定的所述待插值像素点的边缘方向与水平方向的夹角α为0°时，利用双三次插值方式进行插值。
- 如权利要求1所述的图像插值方法，其中，所述沿该边缘方向选取与该待插值像素点相邻的M个原始像素点，以该M个原始像素点作为控制点进行插值计算的步骤具体包括：在该待插值像素点上、下两个相邻原始像素行，沿该边缘方向选取最邻近的4个原始像素点，该待插值像素点的灰度值由该4个原始像素点作为控制点进行双线性插值计算获得。
- 如权利要求1至8中任一权利要求所述的图像插值方法，还包括：在对所述图像进行插值时，判断待插值像素点是否处于纹理区域；当判断到待插值像素点处于纹理区域时，检测该待插值像素点的噪声， 根据该噪声选取对应的滤波器进行插值。
- 如权利要求1至8所述的图像插值方法，还包括：在对所述图像进行插值时，判断待插值像素点是否处于平坦区域；当判断到待插值像素点处于平坦区域时，采用三次卷积插值法进行插值计算。
- 一种图像插值装置，所述图像包括边缘区域、纹理区域和平坦区域；其中，所述图像插值装置包括：判断模块，用于对待插值像素点进行区域判断；以及，插值模块，用于当所述判断模块判断到待插值像素点处于边缘区域时，确定该待插值像素点的边缘方向，沿该边缘方向选取与该待插值像素点相邻的M个原始像素点，以该M个原始像素点作为控制点进行插值计算，M为大于1的整数。
- 如权利要求11所述的图像插值装置，其中，所述判断模块，还用于在对所述图像进行插值时，判断待插值像素点是否处于纹理区域；所述插值模块，还用于当所述判断模块判断到待插值像素点处于纹理区域时，检测该待插值像素点的噪声，根据该噪声选取对应的滤波器进行插值。
- 如权利要求11所述的图像插值装置，其中，所述判断模块，还用于在对所述图像进行插值时，判断待插值像素点是否处于平坦区域；所述插值模块，还用于当所述判断模块判断到待插值像素点处于平坦区域时，采用三次卷积插值法进行插值计算。
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410422383.1A CN104200426A (zh) | 2014-08-25 | 2014-08-25 | 图像插值方法和装置 |
| CN201410422383.1 | 2014-08-25 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2016029555A1 true WO2016029555A1 (zh) | 2016-03-03 |
Family
ID=52085712
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2014/091076 WO2016029555A1 (zh) | 2014-08-25 | 2014-11-14 | 图像插值方法和装置 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN104200426A (zh) |
| WO (1) | WO2016029555A1 (zh) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111476723A (zh) * | 2020-03-17 | 2020-07-31 | 哈尔滨师范大学 | 一种Landsat-7扫描线纠正器失效的遥感图像丢失像素恢复方法 |
| CN113837943A (zh) * | 2021-09-28 | 2021-12-24 | 广州极飞科技股份有限公司 | 图像处理方法、装置、电子设备及可读存储介质 |
| CN113936015A (zh) * | 2021-12-17 | 2022-01-14 | 青岛美迪康数字工程有限公司 | 一种图像有效区域的提取方法及装置 |
| CN116740579A (zh) * | 2023-08-15 | 2023-09-12 | 兰陵县城市规划设计室 | 一种国土空间规划数据智能采集方法 |
| CN117912404A (zh) * | 2024-03-15 | 2024-04-19 | 杭州海康威视数字技术股份有限公司 | 一种应用于led显示屏的灯珠有效像素点的边界确定方法 |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3276937A4 (en) * | 2015-03-25 | 2018-09-05 | Olympus Corporation | Image processing device, image processing method, and program |
| WO2016154966A1 (zh) * | 2015-04-01 | 2016-10-06 | 中国科学院自动化研究所 | 基于边缘自适应的图像缩放方法及系统 |
| CN104881843A (zh) * | 2015-06-10 | 2015-09-02 | 京东方科技集团股份有限公司 | 图像插值方法和图像插值装置 |
| CN106373090B (zh) * | 2016-08-31 | 2019-11-15 | 广州视睿电子科技有限公司 | 图片处理方法及装置 |
| CN106780383B (zh) * | 2016-12-13 | 2019-05-24 | 长春理工大学 | Tof相机的深度图像增强方法 |
| CN109658333A (zh) * | 2018-11-14 | 2019-04-19 | 深圳市华星光电半导体显示技术有限公司 | 图像放大插值的方法、图像放大插值装置以及显示装置 |
| CN116993723B (zh) * | 2023-09-26 | 2023-12-15 | 山东弘密新材料科技有限公司 | 基于图像处理的壁纸质量检测方法 |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1667650A (zh) * | 2005-04-08 | 2005-09-14 | 杭州国芯科技有限公司 | 基于边缘检测的图像缩放的方法 |
| US20080165206A1 (en) * | 2007-01-04 | 2008-07-10 | Himax Technologies Limited | Edge-Oriented Interpolation Method and System for a Digital Image |
| 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 |
| CN103456031A (zh) * | 2013-07-08 | 2013-12-18 | 江南大学 | 一种区域图像插值的新方法 |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6832009B1 (en) * | 1999-09-24 | 2004-12-14 | Zoran Corporation | Method and apparatus for improved image interpolation |
| US6771835B2 (en) * | 2000-06-12 | 2004-08-03 | Samsung Electronics Co., Ltd. | Two-dimensional non-linear interpolation system based on edge information and two-dimensional mixing interpolation system using the same |
-
2014
- 2014-08-25 CN CN201410422383.1A patent/CN104200426A/zh active Pending
- 2014-11-14 WO PCT/CN2014/091076 patent/WO2016029555A1/zh active Application Filing
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1667650A (zh) * | 2005-04-08 | 2005-09-14 | 杭州国芯科技有限公司 | 基于边缘检测的图像缩放的方法 |
| US20080165206A1 (en) * | 2007-01-04 | 2008-07-10 | Himax Technologies Limited | Edge-Oriented Interpolation Method and System for a Digital Image |
| 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 |
| CN103456031A (zh) * | 2013-07-08 | 2013-12-18 | 江南大学 | 一种区域图像插值的新方法 |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111476723A (zh) * | 2020-03-17 | 2020-07-31 | 哈尔滨师范大学 | 一种Landsat-7扫描线纠正器失效的遥感图像丢失像素恢复方法 |
| CN113837943A (zh) * | 2021-09-28 | 2021-12-24 | 广州极飞科技股份有限公司 | 图像处理方法、装置、电子设备及可读存储介质 |
| CN113936015A (zh) * | 2021-12-17 | 2022-01-14 | 青岛美迪康数字工程有限公司 | 一种图像有效区域的提取方法及装置 |
| CN116740579A (zh) * | 2023-08-15 | 2023-09-12 | 兰陵县城市规划设计室 | 一种国土空间规划数据智能采集方法 |
| CN116740579B (zh) * | 2023-08-15 | 2023-10-20 | 兰陵县城市规划设计室 | 一种国土空间规划数据智能采集方法 |
| CN117912404A (zh) * | 2024-03-15 | 2024-04-19 | 杭州海康威视数字技术股份有限公司 | 一种应用于led显示屏的灯珠有效像素点的边界确定方法 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104200426A (zh) | 2014-12-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2016029555A1 (zh) | 图像插值方法和装置 | |
| WO2019205290A1 (zh) | 一种图像检测方法、装置、计算机设备及存储介质 | |
| JP4955096B2 (ja) | 検出装置、検出方法、検出プログラム、および記録媒体 | |
| EP3764318A1 (en) | Systems and methods for three dimensional geometric reconstruction of captured image data | |
| US20190164008A1 (en) | Image processing apparatus, image processing method, and storage medium | |
| US10748023B2 (en) | Region-of-interest detection apparatus, region-of-interest detection method, and recording medium | |
| CN103198455B (zh) | 一种利用全变差最小化和灰度共生矩阵的图像去噪方法 | |
| CN107292828B (zh) | 图像边缘的处理方法和装置 | |
| KR20140000195A (ko) | 입체 카메라를 위한 자동초점 | |
| KR101032446B1 (ko) | 영상의 정점 검출 장치 및 방법 | |
| US9245194B2 (en) | Efficient line detection method | |
| CN113781406B (zh) | 电子元器件的划痕检测方法、装置及计算机设备 | |
| WO2017088637A1 (zh) | 自然背景中图像边缘的定位方法及装置 | |
| US9373053B2 (en) | Image processor with edge selection functionality | |
| JPWO2014033936A1 (ja) | 画像処理装置、画像処理方法および画像処理プログラム | |
| CN108665495B (zh) | 图像处理方法及装置、移动终端 | |
| TWI466063B (zh) | 影像內插的處理方法 | |
| US20150131853A1 (en) | Stereo matching system and method for generating disparity map using same | |
| CN109101874B (zh) | 一种基于深度图像的图书馆机器人障碍识别方法 | |
| CN112465707A (zh) | 红外图像条纹噪声的处理方法、装置、介质及电子设备 | |
| CN106951902B (zh) | 一种图像二值化处理方法及装置 | |
| US9070015B2 (en) | System and method for iris detection in digital images | |
| CN111801709A (zh) | 一种圆形特征检测方法、处理系统及具有存储功能的装置 | |
| TW202008217A (zh) | 比對追蹤物件的方法及電子裝置 | |
| CN111383183B (zh) | 图像边缘增强方法、装置以及计算机存储介质 |
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: 14900715 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 04.07.2017) |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 14900715 Country of ref document: EP Kind code of ref document: A1 |