WO2024066090A1 - Corner detection method and system based on texture features, electronic device, and medium - Google Patents
Corner detection method and system based on texture features, electronic device, and medium Download PDFInfo
- Publication number
- WO2024066090A1 WO2024066090A1 PCT/CN2022/141469 CN2022141469W WO2024066090A1 WO 2024066090 A1 WO2024066090 A1 WO 2024066090A1 CN 2022141469 W CN2022141469 W CN 2022141469W WO 2024066090 A1 WO2024066090 A1 WO 2024066090A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- image
- texture
- corner
- corner detection
- processed
- Prior art date
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 139
- 238000000034 method Methods 0.000 claims abstract description 42
- 238000000605 extraction Methods 0.000 claims abstract description 13
- 238000001914 filtration Methods 0.000 claims description 43
- 230000006870 function Effects 0.000 claims description 12
- 230000004927 fusion Effects 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 5
- 238000003672 processing method Methods 0.000 claims 1
- 238000010586 diagram Methods 0.000 description 7
- 238000012545 processing Methods 0.000 description 5
- 238000005316 response function Methods 0.000 description 4
- 101100233916 Saccharomyces cerevisiae (strain ATCC 204508 / S288c) KAR5 gene Proteins 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 101001121408 Homo sapiens L-amino-acid oxidase Proteins 0.000 description 2
- 102100026388 L-amino-acid oxidase Human genes 0.000 description 2
- 101100012902 Saccharomyces cerevisiae (strain ATCC 204508 / S288c) FIG2 gene Proteins 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 101000827703 Homo sapiens Polyphosphoinositide phosphatase Proteins 0.000 description 1
- 102100023591 Polyphosphoinositide phosphatase Human genes 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/40—Analysis of texture
Definitions
- the present invention discloses a corner point detection method, system, electronic equipment and medium based on texture features.
- Interest points in an image are points in an image that are easy to detect and have representative significance. Based on these points, target recognition, image matching, and defect detection can be performed. Corner points are the most basic type of interest points in an image. They can be defined as the intersection of two edges in an image, or the local maximum point of curvature on the contour line of a target. Corner point features have the characteristics of low computational complexity, simple matching, and invariance to rotation, translation, and scaling. Therefore, they play a very important role in application fields such as image registration and matching, target recognition, motion analysis, and target tracking. At present, image corner point detection is heavily dependent on the order of detection points and the distribution near the corner points. It is difficult to show that the pixel position selected for comparison can best reflect the performance of the corner points; and because all pixels are traversed, it takes a lot of time.
- image corner detection is heavily dependent on the order of detection points and the distribution near the corner points. It is difficult to prove that the pixel position selected for comparison can best reflect the corner performance; and because all pixels are traversed, it takes a lot of time.
- a texture feature-based corner detection method system, electronic device, and medium are provided.
- a corner point detection method based on texture features comprising:
- corner point detection is performed on the fused image.
- the method before extracting texture features of the image to be processed, the method further includes:
- the filtering method includes at least one of the following: Gaussian filtering method and Gabor filtering method.
- using Gaussian filtering to filter the image to be processed includes:
- Gaussian i represents the Gaussian kernel function with the i-th scale factor ⁇ i and radius R i
- I(x,y) in represents the image to be processed
- I(x,y) Ri represents the i-th filtered image.
- texture features are extracted from the filtered image to obtain a texture image, and the formula is as follows:
- I(x,y) Detaili represents the i-th detail image
- I(x,y) Detail represents the texture image
- wi represents the i-th weight value
- the fused image I(x,y) out is:
- I(x,y) out I(x,y) in +I(x,y) Detail .
- the corner detection method includes at least one of the following: SIFT corner detection, ORB corner detection, FAST corner detection, Harris corner detection, and SURF corner detection.
- a corner point detection system based on texture features comprising:
- a texture extraction unit configured to extract texture features from the image to be processed to obtain a texture image
- an image fusion unit configured to fuse the texture image with the image to be processed to obtain a fused image
- the corner point detection unit is configured to perform corner point detection on the fused image based on the texture image.
- system further comprises:
- An image filtering unit is configured to perform filtering on the image to be processed to obtain a filtered image.
- the image filtering unit is further configured to use Gaussian kernel functions with different scale factors and different radii to perform scale processing on the image to be processed to obtain a filtered image.
- the image filtering unit is further configured to use Gaussian kernel functions with different scale factors and different radii to perform scale processing on the image to be processed to obtain a filtered image.
- the texture extraction unit is configured to obtain a plurality of detail images according to the image to be processed and filtered images of different scales, and configure corresponding weights for the plurality of detail images respectively to obtain the texture image.
- the image fusion unit is configured to add the texture image and the image to be processed to obtain a fused image.
- the corner detection unit is configured to perform at least one of the following methods: SIFT corner detection, ORB corner detection, FAST corner detection, Harris corner detection, and SURF corner detection.
- An electronic device includes a memory and one or more processors, wherein the memory stores computer-readable instructions, and when the one or more processors execute the computer-readable instructions, the steps of the texture feature-based corner detection method provided in any embodiment of the present disclosure are implemented.
- One or more non-volatile computer-readable storage media storing computer-readable instructions, on which computer-readable instructions are stored, and when the computer-readable instructions are executed by one or more processors, the steps of the corner point detection method based on texture features provided in any embodiment of the present disclosure are implemented.
- FIG1 is an exemplary flow chart of a method for detecting corner points based on texture features according to one or more embodiments of the present disclosure
- FIG2 is another exemplary flow chart of a corner point detection method based on texture features provided by one or more embodiments of the present disclosure
- FIG3 is a comparison diagram of an image to be processed and an image after fusion provided by one or more embodiments of the present disclosure; wherein FIG3 (a) is an image to be processed; and FIG3 (b) is an image after fusion;
- FIG4 is an exemplary flow chart of a FAST corner point detection method provided by one or more embodiments of the present disclosure
- FIG5 is an exemplary structural block diagram of a corner point detection system based on texture features provided by one or more embodiments of the present disclosure
- FIG6 is another exemplary structural block diagram of a corner point detection system based on texture features provided by one or more embodiments of the present disclosure
- FIG7 is a schematic diagram of the structure of an electronic device provided by one or more embodiments of the present disclosure.
- an embodiment of the present disclosure provides a corner detection method 100 based on texture features, comprising:
- S140 Based on the texture image, perform corner point detection on the fused image.
- the image to be processed is any frame image extracted from the acquired video, and the frame image is used as the original image for texture feature extraction to separate the background and texture of the original image, thereby obtaining a texture image, and the texture image is fused with the image to be processed to obtain a fused image. Corner point detection is performed on the fused image based on the texture image, which can greatly save computing power and time, and image registration can also be performed based on the texture image later.
- the corner point detection method based on texture features provided by the embodiments of the present disclosure is independent of the order of detection points and the distribution near the corner points. Since the corner points are basically located at the texture features, the selected pixel positions for performing corner point detection based on the texture features of the image can better reflect the performance of the corner points, with short detection time and high detection efficiency and accuracy.
- step S120 the following steps are also included:
- the image to be processed is first filtered, and then texture feature extraction is performed on the filtered image. Filtering the image to be processed can remove noise in the image, avoid mistaking noise points in the image for corner points, and improve the accuracy of corner point detection.
- the filtering method includes at least one of the following: Gaussian filtering method and Gabor filtering method.
- a Gaussian filter method may be used to remove noise in the image to be processed, thereby improving the accuracy of subsequent image corner point detection.
- a Gabor filter method may be used to remove noise in the image to be processed, thereby improving the accuracy of subsequent image corner point detection.
- step S110 filtering the image to be processed using a Gaussian filtering method includes:
- i represents the number of filtering times
- ⁇ i represents the i-th scale factor
- R i represents the i-th radius
- Gaussian i represents the Gaussian kernel function with the i-th scale factor ⁇ i and radius R i
- I(x,y) in represents the image to be processed
- I(x,y) Ri represents the i-th filtered image.
- the Gaussian kernel function is a method for constructing a scale space for an image in image processing, which uses Gaussian kernels Gaussian i with different scale factors ⁇ i and different radii R i to convolve the image, so as to scale the image and obtain filtered images of different scales.
- Gaussian kernels Gaussian i with different scale factors ⁇ i and different radii R i to convolve the image, so as to scale the image and obtain filtered images of different scales.
- the image to be processed is subjected to three Gaussian filterings, as follows:
- step S120 texture features are extracted from the filtered image to obtain a texture image, and the formula is as follows:
- I(x,y) Detaili represents the i-th detail image
- I(x,y) Detail represents the texture image
- wi represents the i-th weight value
- a plurality of detail images are first obtained according to the image to be processed and the filtered images of different scales, and then corresponding weights are respectively configured for the plurality of detail images to obtain the texture image.
- the weight value wi is not particularly limited, and those skilled in the art can arbitrarily select and set it according to actual needs.
- I(x,y) Detail2 I(x,y) R1 - I(x,y) R2
- I(x,y) Detail (1- w1 ⁇ sgn(I(x,y) Detail1 )) ⁇ I(x,y) Detail1 + w2 ⁇ I(x,y) Detail2 + w3 ⁇ I(x,y) Detail3
- step S130 the fused image I(x, y) out is:
- I(x,y) out I(x,y) in + I(x,y) Detail
- Figure 3(a) shows the image to be processed, i.e., the original image
- Figure 3(b) shows the fused image.
- the corner detection method includes at least one of the following: SIFT corner detection, ORB corner detection, FAST corner detection, Harris corner detection, and SURF corner detection.
- corner detection can be performed based on the texture image in conjunction with a variety of corner detection methods, such as SIFT (Scale-invariant feature transform) corner detection, ORB (Oriented FAST and Rotated BRIEF) corner detection, FAST (Features from Accelerated Segment Test) corner detection, Harris corner detection, SURF (Speeded Up Robust Features) corner detection and any other corner detection method.
- SIFT Scale-invariant feature transform
- ORB Oriented FAST and Rotated BRIEF corner detection
- FAST Features from Accelerated Segment Test
- Harris corner detection Harris corner detection
- SURF Speeded Up Robust Features
- Harris corner detection is the point with sufficiently high grayscale change value in all directions within the neighborhood, and it is the point with maximum curvature on the edge curve of the image.
- the basic idea of corner detection is: use the corner detection operator to calculate the corner response function (Corner Response Function) for each pixel of the image, threshold the corner response function, select the threshold according to the actual situation, perform non-maximum suppression on the thresholded corner response function, and obtain non-zero points as corner points; detect corner points in the neighborhood through a small sliding window and move the window in any direction. If the grayscale value in the window changes dramatically, the center of the window is the corner point.
- a FAST corner detection method is used to perform corner detection on the fused image, including:
- a point P'(x, y) is determined on the texture image, and the point P(x, y) corresponding to the point P'(x, y) is found in the fused image. Then, the fused image is subjected to corner detection. For example, based on the FAST corner detection method, the point P(x, y) is judged with the 16 pixels on the circle around the point to determine whether the point P(x, y) is a corner point, that is, whether the point P'(x, y) is a corner point. Then, the next point in the texture image is determined to be a corner point in turn, until all the corner points on the texture image are found, and the corner detection is terminated. Since all corner points are basically located in the texture, by checking whether the texture point is a corner point, the detection time can be shortened and the detection efficiency can be improved.
- the FAST corner detection method is used for corner detection.
- the FAST method is a corner detection method based on templates and machine learning. It not only has fast calculation speed but also has high accuracy.
- the FAST method mainly considers 16 pixels on the circular window near the pixel point, as shown in Figure 4 below. p is the center pixel point, and the point pixel marked by the white box is the point we need to consider.
- the FAST corner point is defined as: if a certain number of pixels around a pixel have different pixel values from the point, it is considered a corner point.
- the specific corner point detection includes the following steps:
- the threshold t has different values in different scenes. If the threshold t is 0, it can be understood that there are Q consecutive pixels greater than or less than the grayscale value of IP , then the point is a corner point. In the disclosed embodiment, Q is usually 12 or 9. Experiments show that the corner point detection performance is most stable, faster, and has good results when Q is 12. There are also experiments showing that the corner point detection performance is better when Q is 9.
- the specific corner point detection formula is as follows:
- IP represents the pixel value of the center point P
- IP ⁇ x represents the pixel point at the circular template x around the point P
- the pixel belongs to darker
- Sp ⁇ x d
- the other two cases represent brighter and similar, respectively. Therefore, the circular area around the center point P is divided into three types: d, s and b. If the number of occurrences of d or b is greater than Q (Q is 12 or 9), the point is considered to be a candidate corner point.
- an improved FAST corner detection method is used to perform corner detection on the fused image, including:
- Set a threshold t select four points in mutually perpendicular directions from the M pixels on the circle around point P(x,y), if at least three of the four points have pixel values less than IP -t or greater than IP +t, then point P(x,y) is a candidate corner point, repeat the judgment until there are Q consecutive pixels less than IP -t or greater than IP +t among the M pixels on the circle around point P(x,y), then point P(x,y) is a corner point; M ⁇ Q.
- the size of the pixel value of the four points in the horizontal and vertical directions on the circle around the center point P can be compared with the center point, such as the pixel values of points 1, 5, 9, and 13.
- the center point such as the pixel values of points 1, 5, 9, and 13.
- 2 may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.
- a corner point detection system 200 based on texture features in this embodiment includes:
- the texture extraction unit 220 is configured to extract texture features from the image to be processed to obtain a texture image
- An image fusion unit 230 configured to fuse the texture image with the image to be processed to obtain a fused image
- the corner point detection unit 240 is configured to perform corner point detection on the fused image based on the texture image.
- the system separates the background and texture of the image to be processed to obtain a texture image, performs corner point detection based on the texture features of the image, and determines whether the texture features are corner points.
- the selected pixel positions can better reflect the performance of the corner points, which can greatly save computing power and time.
- the system further includes:
- the image filtering unit 210 is configured to perform filtering on the image to be processed to obtain a filtered image.
- the image filtering unit 210 can filter the image to be processed to remove noise in the image, avoid mistaking noise points in the image for corner points, and improve the accuracy of corner point detection.
- the image filtering unit 210 can be configured to perform Gaussian filtering or Gabor filtering.
- the image filtering unit 210 is further configured to use Gaussian kernel functions with different scale factors and different radii to perform scale processing on the image to be processed to obtain a filtered image.
- the image filtering unit 210 uses a Gaussian filtering method to filter the image to be processed.
- the specific Gaussian filtering method is as described above, and the embodiments of the present disclosure will not be described in detail.
- the texture extraction unit 220 is configured to obtain a plurality of detail images according to the image to be processed and filtered images of different scales, and configure corresponding weights for the plurality of detail images respectively to obtain a texture image.
- the texture extraction unit 220 is configured to process the image to be processed and the filtered images of different scales to obtain a texture image.
- the specific method for extracting the texture image is as described above.
- the image fusion unit 230 is configured to add the texture image and the image to be processed to obtain a fused image.
- the corner detection unit 240 is configured to perform at least one of the following methods: SIFT corner detection, ORB corner detection, FAST corner detection, Harris corner detection, and SURF corner detection.
- the detection method performed by the corner point detection unit 240 is as described above, and the embodiments of the present disclosure will not be described in detail.
- an electronic device in a third aspect of the embodiments of the present disclosure, is provided, and its internal structure diagram may be shown in FIG7.
- the electronic device includes one or more processors, memory, communication interface, display screen and input device connected via a system bus. Among them, the one or more processors of the electronic device are configured to provide computing and control capabilities.
- the memory of the electronic device includes a non-volatile storage medium and an internal memory.
- the non-volatile storage medium stores an operating system and computer-readable instructions.
- the internal memory provides an environment for the operation of the operating system and computer-readable instructions in the non-volatile storage medium.
- the communication interface of the electronic device is configured to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be implemented through WIFI, an operator network, near field communication (NFC) or other technologies.
- WIFI wireless fidelity
- NFC near field communication
- the computer-readable instructions are executed by one or more processors, a corner detection method based on texture features is implemented.
- the display screen of the electronic device may be a liquid crystal display screen or an electronic ink display screen
- the input device of the electronic device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the electronic device, or an external keyboard, touchpad or mouse, etc.
- FIG. 7 is merely a block diagram of a partial structure related to the scheme of the present disclosure, and does not constitute a limitation on the electronic device to which the scheme of the present disclosure is applied.
- the specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.
- the texture feature-based corner detection system 200 provided by the present disclosure may be implemented in the form of a computer-readable instruction, and the computer-readable instruction may be run on an electronic device as shown in FIG7.
- the memory of the electronic device may store various program modules constituting the texture feature-based corner detection system 200, such as the image filtering unit 210, the texture extraction unit 220, the image fusion unit 230, the corner detection unit 240, etc. shown in FIG6.
- the computer-readable instructions constituted by the various program modules enable one or more processors to execute the steps of the texture feature-based corner detection method of various embodiments of the present disclosure described in this specification.
- an electronic device including a memory and one or more processors, wherein the memory stores computer-readable instructions, and the one or more processors implement the following steps of the above-mentioned texture feature-based corner detection method when executing the computer-readable instructions.
- one or more non-volatile computer-readable storage media storing computer-readable instructions are also provided, on which computer-readable instructions are stored.
- the steps in the above-mentioned method embodiments are implemented.
- Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc.
- Volatile memory can include random access memory (RAM) or external cache memory.
- RAM is available in many forms, such as static random access memory (SRAM) and dynamic random access memory (DRAM).
- the present invention provides a corner point detection method based on texture features.
- the detection of corner points does not depend on the order of detection points and the distribution near the corner points. Since the corner points are basically located at the texture features, the corner point detection is performed based on the texture features of the image. The selected pixel positions can better reflect the performance of the corner points.
- the detection time is short, the detection efficiency and accuracy are high, and the method has strong industrial applicability.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The present disclosure relates to the technical field of image detection, and provided thereby are a corner detection method and system based on texture features, an electronic device, and a medium. The method comprises: performing texture feature extraction on an image to be processed, and obtaining a texture image; fusing the texture image and the image to be processed, and obtaining a fused image; and on the basis of the texture image, performing corner detection on the fused image. In the method, the background and texture of the image to be processed are separated, the texture image is obtained, the texture image is fused with the image to be processed, the fused image is obtained, and corner detection is performed on the fused image on the basis of the texture image. By means of performing corner detection on texture features of an image using the present method, a selected pixel position can better reflect the properties of a corner, thereby greatly reducing computing power and time consumed.
Description
本公开要求于2022年09月26日提交中国专利局、申请号为2022111782365、发明名称为“基于纹理特征的角点检测方法、***、电子设备及介质”的中国专利申请的优先权,其全部内容通过引用结合在本公开中。This disclosure claims the priority of the Chinese patent application filed with the China Patent Office on September 26, 2022, with application number 2022111782365 and invention name “Corner detection method, system, electronic device and medium based on texture features”, the entire contents of which are incorporated by reference in this disclosure.
本公开一种基于纹理特征的角点检测方法、***、电子设备及介质。The present invention discloses a corner point detection method, system, electronic equipment and medium based on texture features.
图像中的兴趣点(又称为关键点、特征点)是在图像中容易检测且具有代表意义的点。基于这些点,可以进行目标识别、图像匹配和缺陷检测等。角点是图像中最基本的一种兴趣点,它可定义为图像中两个边缘的交点,或目标轮廓线上曲率的局部极大点。角点特征具有计算量少,匹配简单以及旋转、平移、放缩不变性等特点,因此在图像配准与匹配、目标识别、运动分析、目标跟踪等应用领域都起着非常重要的作用。目前,图像角点检测严重依赖于检测点的顺序和角点附近的分布,很难说明所选择比较的像素位置能最好的反应角点性能;且由于遍历所有的像素点,导致耗时较多。Interest points in an image (also known as key points or feature points) are points in an image that are easy to detect and have representative significance. Based on these points, target recognition, image matching, and defect detection can be performed. Corner points are the most basic type of interest points in an image. They can be defined as the intersection of two edges in an image, or the local maximum point of curvature on the contour line of a target. Corner point features have the characteristics of low computational complexity, simple matching, and invariance to rotation, translation, and scaling. Therefore, they play a very important role in application fields such as image registration and matching, target recognition, motion analysis, and target tracking. At present, image corner point detection is heavily dependent on the order of detection points and the distribution near the corner points. It is difficult to show that the pixel position selected for comparison can best reflect the performance of the corner points; and because all pixels are traversed, it takes a lot of time.
发明内容Summary of the invention
(一)要解决的技术问题1. Technical issues to be resolved
目前,图像角点检测严重依赖于检测点的顺序和角点附近的分布,很难说明所选择比较的像素位置能最好的反应角点性能;且由于遍历所有的像素点,导致耗时较多。At present, image corner detection is heavily dependent on the order of detection points and the distribution near the corner points. It is difficult to prove that the pixel position selected for comparison can best reflect the corner performance; and because all pixels are traversed, it takes a lot of time.
(二)技术方案(II) Technical solution
根据本公开公开的各种实施例,提供一种基于纹理特征的角点检测方法、***、电子设备及介质。According to various embodiments disclosed in the present disclosure, a texture feature-based corner detection method, system, electronic device, and medium are provided.
一种基于纹理特征的角点检测方法,包括:A corner point detection method based on texture features, comprising:
对待处理图像进行纹理特征提取,得到纹理图像;Extract texture features of the image to be processed to obtain a texture image;
将所述纹理图像与所述待处理图像进行融合,得到融合后的图像;Fusing the texture image with the image to be processed to obtain a fused image;
基于所述纹理图像,对所述融合后的图像进行角点检测。Based on the texture image, corner point detection is performed on the fused image.
在一个实施例中,所述对待处理图像进行纹理特征提取之前还包括:In one embodiment, before extracting texture features of the image to be processed, the method further includes:
对所述待处理图像进行滤波处理,得到滤波后的图像。Perform filtering on the image to be processed to obtain a filtered image.
在一个实施例中,所述滤波处理的方法包括至少如下一种:高斯滤波法、Gabor滤波法。In one embodiment, the filtering method includes at least one of the following: Gaussian filtering method and Gabor filtering method.
在一个实施例中,采用高斯滤波法对所述待处理图像进行滤波处理包括:In one embodiment, using Gaussian filtering to filter the image to be processed includes:
采用不同尺度因子、不同的半径的高斯核函数对待处理图像进行尺度处理,得到滤波后的图像:Use Gaussian kernel functions with different scale factors and different radii to scale the image to be processed and obtain the filtered image:
I(x,y)
Ri=Gaussian
i(I(x,y)
in);i=[1,2,…,N]
I(x,y) Ri = Gaussian i (I(x,y) in ); i = [1,2,…,N]
其中,i表示滤波次数,Gaussian
i表示第i个尺度因子为σ
i、半径为R
i的高斯核函数,I(x,y)
in表示待处理图像,I(x,y)
Ri表示第i个滤波后的图像。
Where i represents the number of filtering times, Gaussian i represents the Gaussian kernel function with the i-th scale factor σ i and radius R i , I(x,y) in represents the image to be processed, and I(x,y) Ri represents the i-th filtered image.
在一个实施例中,对所述滤波后的图像进行纹理特征提取,得到纹理图像,公式如下:In one embodiment, texture features are extracted from the filtered image to obtain a texture image, and the formula is as follows:
其中,I(x,y)
Detaili表示第i个细节图像;I(x,y)
Detail表示纹理图像,w
i表示第i个权重值。
Among them, I(x,y) Detaili represents the i-th detail image; I(x,y) Detail represents the texture image, and wi represents the i-th weight value.
在一个实施例中,所述融合后的图像I(x,y)
out为:
In one embodiment, the fused image I(x,y) out is:
I(x,y)
out=I(x,y)
in+I(x,y)
Detail。
I(x,y) out =I(x,y) in +I(x,y) Detail .
在一个实施例中,所述角点检测的方法包括至少如下一种:SIFT角点检测、ORB角点检测、FAST角点检测、Harris角点检测、SURF角点检测。In one embodiment, the corner detection method includes at least one of the following: SIFT corner detection, ORB corner detection, FAST corner detection, Harris corner detection, and SURF corner detection.
一种基于纹理特征的角点检测***,包括:A corner point detection system based on texture features, comprising:
纹理提取单元,配置成对待处理图像进行纹理特征提取,得到纹理图像;A texture extraction unit configured to extract texture features from the image to be processed to obtain a texture image;
图像融合单元,配置成将所述纹理图像与所述待处理图像进行融合,得到融合后的图像;an image fusion unit, configured to fuse the texture image with the image to be processed to obtain a fused image;
角点检测单元,配置成基于所述纹理图像,对所述融合后的图像进行角点检测。The corner point detection unit is configured to perform corner point detection on the fused image based on the texture image.
在一个实施例中,所述***还包括:In one embodiment, the system further comprises:
图像滤波单元,所述图像滤波单元配置成对所述待处理图像进行滤波处理,得到滤波后的图像。An image filtering unit is configured to perform filtering on the image to be processed to obtain a filtered image.
在一个实施例中,所述图像滤波单元,还配置成采用不同尺度因子、不同的半径的高斯核函数对待处理图像进行尺度处理,得到滤波后的图像。In one embodiment, the image filtering unit is further configured to use Gaussian kernel functions with different scale factors and different radii to perform scale processing on the image to be processed to obtain a filtered image.
在一个实施例中,所述图像滤波单元,还配置成采用不同尺度因子、不同的半径的高斯核函数对待处理图像进行尺度处理,得到滤波后的图像。In one embodiment, the image filtering unit is further configured to use Gaussian kernel functions with different scale factors and different radii to perform scale processing on the image to be processed to obtain a filtered image.
在一个实施例中,所述纹理提取单元,配置成根据待处理图像以及不同尺度的滤波后的图像得到多个细节图像,并为多个细节图像分别配置相应的权重以获取纹理图像。In one embodiment, the texture extraction unit is configured to obtain a plurality of detail images according to the image to be processed and filtered images of different scales, and configure corresponding weights for the plurality of detail images respectively to obtain the texture image.
在一个实施例中,所述图像融合单元,配置成将所述纹理图像与所述待处理图像进行相加,得到融合后的图像。In one embodiment, the image fusion unit is configured to add the texture image and the image to be processed to obtain a fused image.
在一个实施例中,所述角点检测单元,配置成执行如下至少一种方法:SIFT角点检测、ORB角点检测、FAST角点检测、Harris角点检测、SURF角点检测。In one embodiment, the corner detection unit is configured to perform at least one of the following methods: SIFT corner detection, ORB corner detection, FAST corner detection, Harris corner detection, and SURF corner detection.
一种电子设备,包括存储器和一个或多个处理器,所述存储器存储有计算机可读指令,所述一个或多个处理器执行所述计算机可 读指令时实现本公开任意实施例所提供的基于纹理特征的角点检测方法的步骤。An electronic device includes a memory and one or more processors, wherein the memory stores computer-readable instructions, and when the one or more processors execute the computer-readable instructions, the steps of the texture feature-based corner detection method provided in any embodiment of the present disclosure are implemented.
一个或多个存储有计算机可读指令的非易失性计算机可读存储介质,其上存储有计算机可读指令,所述计算机可读指令被一个或多个处理器执行时实现本公开任意实施例所提供的基于纹理特征的角点检测方法的步骤。One or more non-volatile computer-readable storage media storing computer-readable instructions, on which computer-readable instructions are stored, and when the computer-readable instructions are executed by one or more processors, the steps of the corner point detection method based on texture features provided in any embodiment of the present disclosure are implemented.
本公开的其他特征和优点将在随后的说明书中阐述,并且,部分地从说明书中变得显而易见,或者通过实施本公开而了解。本公开的目的和其他优点在说明书、权利要求书以及附图中所特别指出的结构来实现和获得,本公开的一个或多个实施例的细节在下面的附图和描述中提出。Other features and advantages of the present disclosure will be described in the subsequent description, and partly become apparent from the description, or be understood by practicing the present disclosure. The purpose and other advantages of the present disclosure are realized and obtained by the structures particularly pointed out in the description, claims and drawings, and the details of one or more embodiments of the present disclosure are presented in the following drawings and descriptions.
为使本公开的上述目的、特征和优点能更明显易懂,下文特举可选实施例,并配合所附附图,作详细说明如下。In order to make the above-mentioned objectives, features and advantages of the present disclosure more obvious and understandable, optional embodiments are specifically listed below and described in detail with reference to the attached drawings.
通过阅读参照以下附图所作的对非限制性实施例所作的详细描述,本公开的其它特征、目的和优点将会变得更明显:Other features, objects and advantages of the present disclosure will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:
图1为本公开一个或多个实施例提供的一种基于纹理特征的角点检测方法中的一种示例性流程图;FIG1 is an exemplary flow chart of a method for detecting corner points based on texture features according to one or more embodiments of the present disclosure;
图2为本公开一个或多个实施例提供的一种基于纹理特征的角点检测方法中的另一种示例性流程图;FIG2 is another exemplary flow chart of a corner point detection method based on texture features provided by one or more embodiments of the present disclosure;
图3为本公开一个或多个实施例提供的待处理图像和融合后的图像的对比图;其中,图(a)为待处理图像;图(b)为融合后的图像;FIG3 is a comparison diagram of an image to be processed and an image after fusion provided by one or more embodiments of the present disclosure; wherein FIG3 (a) is an image to be processed; and FIG3 (b) is an image after fusion;
图4为本公开一个或多个实施例提供的FAST角点检测方法的示例性流程图;FIG4 is an exemplary flow chart of a FAST corner point detection method provided by one or more embodiments of the present disclosure;
图5为本公开一个或多个实施例提供的一种基于纹理特征的角点检测***的一种示例性结构框图;FIG5 is an exemplary structural block diagram of a corner point detection system based on texture features provided by one or more embodiments of the present disclosure;
图6为本公开一个或多个实施例提供的一种基于纹理特征的角点检测***的另一种示例性结构框图;FIG6 is another exemplary structural block diagram of a corner point detection system based on texture features provided by one or more embodiments of the present disclosure;
图7为本公开一个或多个实施例提供的一种电子设备的结构示 意图。FIG7 is a schematic diagram of the structure of an electronic device provided by one or more embodiments of the present disclosure.
下面结合附图和实施例对本公开作进一步的详细说明。可以理解的是,此处所描述的具体实施例仅仅用于解释相关发明,而非对该发明的限定。另外还需要说明的是,为了便于描述,附图中仅示出了与发明相关的部分。The present disclosure is further described in detail below in conjunction with the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the invention are shown in the accompanying drawings.
需要说明的是,在不冲突的情况下,本公开中的实施例及实施例中的特征可以相互组合。下面将参考附图并结合实施例来详细说明本公开。It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present disclosure may be combined with each other. The present disclosure will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.
参考图1所示的基于纹理特征的角点检测方法的示例性流程图,本公开实施例提供一种基于纹理特征的角点检测方法100,包括:Referring to the exemplary flow chart of a corner detection method based on texture features shown in FIG1 , an embodiment of the present disclosure provides a corner detection method 100 based on texture features, comprising:
S120:对待处理图像进行纹理特征提取,得到纹理图像;S120: extracting texture features of the image to be processed to obtain a texture image;
S130:将所述纹理图像与所述待处理图像进行融合,得到融合后的图像;S130: Fusing the texture image with the image to be processed to obtain a fused image;
S140:基于所述纹理图像,对所述融合后的图像进行角点检测。S140: Based on the texture image, perform corner point detection on the fused image.
具体的,待处理图像为从获取的视频中提取的任意一帧图像,将该帧图像作为原始图像进行纹理特征提取,使原始图像的背景与纹理进行分离,从而得到纹理图像,并将纹理图像与待处理图像融合,得到融合后的图像,基于纹理图像对融合后的图像进行角点检测,可以大大节省算力和耗时,且后续也可以基于纹理图像进行图像的配准。Specifically, the image to be processed is any frame image extracted from the acquired video, and the frame image is used as the original image for texture feature extraction to separate the background and texture of the original image, thereby obtaining a texture image, and the texture image is fused with the image to be processed to obtain a fused image. Corner point detection is performed on the fused image based on the texture image, which can greatly save computing power and time, and image registration can also be performed based on the texture image later.
本公开实施例提供的基于纹理特征的角点检测方法,角点的检测不依赖于检测点的顺序和角点附近的分布,由于角点基本位于纹理特征处,因此,基于图像的纹理特征进行角点检测,所选择的像素位置能更好的反应角点的性能,检测时间短,检测效率和准确性高。The corner point detection method based on texture features provided by the embodiments of the present disclosure is independent of the order of detection points and the distribution near the corner points. Since the corner points are basically located at the texture features, the selected pixel positions for performing corner point detection based on the texture features of the image can better reflect the performance of the corner points, with short detection time and high detection efficiency and accuracy.
在一个实施例中,在步骤S120之前还包括以下步骤:In one embodiment, before step S120, the following steps are also included:
S110:对所述待处理图像进行滤波处理,得到滤波后的图像。S110: performing filtering processing on the image to be processed to obtain a filtered image.
具体的,如图2所示,在对待处理图像进行纹理特征提取之前, 先对待处理图像进行滤波处理,再对滤波后的图像进行纹理特征提取,过对待处理图像进行滤波处理可以去除图像中的噪声,避免将图像中的噪声点被误认为角点,提高角点检测的准确性。Specifically, as shown in FIG2 , before texture feature extraction is performed on the image to be processed, the image to be processed is first filtered, and then texture feature extraction is performed on the filtered image. Filtering the image to be processed can remove noise in the image, avoid mistaking noise points in the image for corner points, and improve the accuracy of corner point detection.
在一个实施例中,步骤S110中,所述滤波处理的方法包括至少如下一种:高斯滤波法、Gabor滤波法。In one embodiment, in step S110, the filtering method includes at least one of the following: Gaussian filtering method and Gabor filtering method.
具体的,可以采用高斯滤波法、Gabor滤波法等对待处理图像中的噪声进行去除,提高后续图像角点检测的准确性。Specifically, a Gaussian filter method, a Gabor filter method, or the like may be used to remove noise in the image to be processed, thereby improving the accuracy of subsequent image corner point detection.
在一个实施例中,步骤S110中,采用高斯滤波法对所述待处理图像进行滤波处理包括:In one embodiment, in step S110, filtering the image to be processed using a Gaussian filtering method includes:
采用不同尺度因子、不同的半径的高斯核函数对待处理图像进行尺度处理,得到滤波后的图像:Use Gaussian kernel functions with different scale factors and different radii to scale the image to be processed and obtain the filtered image:
I(x,y)
Ri=Gaussian
i(I(x,y)
in);i=[1,2,…,N]
I(x,y) Ri = Gaussian i (I(x,y) in ); i = [1,2,…,N]
其中,i表示滤波次数,σ
i表示第i个尺度因子,R
i表示第i个半径,Gaussian
i表示第i个尺度因子为σ
i、半径为R
i的高斯核函数,I(x,y)
in表示待处理图像,I(x,y)
Ri表示第i个滤波后的图像。
Wherein, i represents the number of filtering times, σ i represents the i-th scale factor, R i represents the i-th radius, Gaussian i represents the Gaussian kernel function with the i-th scale factor σ i and radius R i , I(x,y) in represents the image to be processed, and I(x,y) Ri represents the i-th filtered image.
具体的,高斯核函数是图像处理中对图像构造尺度空间的方法,是使用不同尺度因子σ
i、不同半径R
i的高斯核Gaussian
i对图像进行卷积,已达到对图像进行尺度处理,得到不同尺度的滤波后的图像。本公开实施例中,对滤波次数i、尺度因子σ
i、半径R
i的具体数值不做特别限制,本领域的技术人员可以根据实际需求任意选择。
Specifically, the Gaussian kernel function is a method for constructing a scale space for an image in image processing, which uses Gaussian kernels Gaussian i with different scale factors σ i and different radii R i to convolve the image, so as to scale the image and obtain filtered images of different scales. In the disclosed embodiment, there is no particular restriction on the specific values of the number of filtering times i, the scale factor σ i , and the radius R i , and those skilled in the art can select them arbitrarily according to actual needs.
示例性的,如滤波次数i的最大值N取3,即对待处理图像进行三次高斯滤波,具体如下:For example, if the maximum value N of the number of filtering times i is 3, the image to be processed is subjected to three Gaussian filterings, as follows:
I(x,y)
R1=Gaussian
1(I(x,y)
in),σ
1=1.0,R
1=r
I(x,y) R1 = Gaussian 1 (I(x,y) in ),σ 1 = 1.0, R 1 = r
I(x,y)
R2=Gaussian
2(I(x,y)
in),σ
2=2.0,R
1=2r
I(x,y) R2 = Gaussian 2 (I(x,y) in ),σ 2 = 2.0, R 1 = 2r
I(x,y)
R3=Gaussian
3(I(x,y)
in),σ
3=4.0,R
1=4r
I(x,y) R3 = Gaussian 3 (I(x,y) in ),σ 3 = 4.0, R 1 = 4r
通过采用三个不同的尺度因子σ
1、σ
2、σ
3对待处理图像分别进行处理,得到三个不同尺度的滤波后的图像,便于后续根据不同尺度的滤波后的图像提取图像的细节信息以得到相应的纹理特征图。
By using three different scale factors σ 1 , σ 2 , σ 3 to process the image to be processed respectively, three filtered images of different scales are obtained, which facilitates the subsequent extraction of image detail information according to the filtered images of different scales to obtain the corresponding texture feature map.
在一个实施例中,步骤S120中,对所述滤波后的图像进行纹理特征提取,得到纹理图像,公式如下:In one embodiment, in step S120, texture features are extracted from the filtered image to obtain a texture image, and the formula is as follows:
其中,I(x,y)
Detaili表示第i个细节图像;I(x,y)
Detail表示纹理图像,w
i表示第i个权重值。
Among them, I(x,y) Detaili represents the i-th detail image; I(x,y) Detail represents the texture image, and wi represents the i-th weight value.
具体的,先根据待处理图像以及不同尺度的滤波后的图像得到多个细节图像,再为多个细节图像分别配置相应的权重以获取纹理图像。本公开实施例中,对权重值w
i不做特别限定,本领域的技术人员可以根据实际需求任意选择设置。
Specifically, a plurality of detail images are first obtained according to the image to be processed and the filtered images of different scales, and then corresponding weights are respectively configured for the plurality of detail images to obtain the texture image. In the disclosed embodiment, the weight value wi is not particularly limited, and those skilled in the art can arbitrarily select and set it according to actual needs.
示例性的,如权重值w
1=0.50,w
2=0.50,w
3=0.25,则多个细节图像和纹理图像如下:
For example, if the weight values w 1 =0.50, w 2 =0.50, and w 3 =0.25, then the multiple detail images and texture images are as follows:
I(x,y)
Detail1=I(x,y)
in-I(x,y)
R1
I(x,y) Detail1 =I(x,y) in -I(x,y) R1
I(x,y)
Detail2=I(x,y)
R1-I(x,y)
R2
I(x,y) Detail2 = I(x,y) R1 - I(x,y) R2
I(x,y)
Detail3=I(x,y)
R2-I(x,y)
R3
I(x,y) Detail3 =I(x,y) R2 -I(x,y) R3
I(x,y)
Detail=(1-w
1×sgn(I(x,y)
Detail1))×I(x,y)
Detail1+w
2×I(x,y)
Detail2+w
3×I(x,y)
Detail3
I(x,y) Detail =(1- w1 ×sgn(I(x,y) Detail1 ))×I(x,y) Detail1 + w2 ×I(x,y) Detail2 + w3 ×I(x,y) Detail3
在一个实施例中,步骤S130中,所述融合后的图像I(x,y)
out为:
In one embodiment, in step S130, the fused image I(x, y) out is:
I(x,y)
out=I(x,y)
in+I(x,y)
Detail
I(x,y) out = I(x,y) in + I(x,y) Detail
具体的,将待处理的图像和纹理图像相加,得到融合后的图像。图3(a)待处理的图像即原始图像,图3(b)为融合后的图像,通过对比图3(a)和图3(b)可知,融合后的图像的纹理特征更加明显,有利于角点的检测。Specifically, the image to be processed and the texture image are added to obtain a fused image. Figure 3(a) shows the image to be processed, i.e., the original image, and Figure 3(b) shows the fused image. By comparing Figure 3(a) and Figure 3(b), it can be seen that the texture features of the fused image are more obvious, which is conducive to the detection of corner points.
在一个实施例中,S140中的,所述角点检测的方法包括至少如下一种:SIFT角点检测、ORB角点检测、FAST角点检测、Harris角点检测、SURF角点检测。In one embodiment, in S140, the corner detection method includes at least one of the following: SIFT corner detection, ORB corner detection, FAST corner detection, Harris corner detection, and SURF corner detection.
具体的,本公开实施例中,基于纹理图像,可以配合多种角点检测方法进行角点检测,如SIFT(Scale-invariant feature transform, 尺度不变性特征变换)角点检测、ORB(Oriented FAST and Rotated BRIEF)角点检测、FAST(Features from Accelerated Segment Test)角点检测、Harris角点检测、SURF(Speeded Up Robust Features,加速鲁棒性特征)角点检测等任意一种角点检测方法。Specifically, in the embodiments of the present disclosure, corner detection can be performed based on the texture image in conjunction with a variety of corner detection methods, such as SIFT (Scale-invariant feature transform) corner detection, ORB (Oriented FAST and Rotated BRIEF) corner detection, FAST (Features from Accelerated Segment Test) corner detection, Harris corner detection, SURF (Speeded Up Robust Features) corner detection and any other corner detection method.
其中,Harris角点检测是在邻域内的各个方向上灰度变化值足够高的点,是图像边缘曲线上曲率极大值的点,角点检测的基本思想是:使用角点检测算子,对图像的每个像素计算角点响应函数(Corner Response Function),阈值化角点响应函数,根据实际情况选择阈值,对阈值化的角点响应函数进行非极大值抑制,并获取非零点作为角点;通过一个小的滑动窗口在邻域检测角点在任意方向上移动窗口,若窗口内的灰度值都有剧烈的变化,则窗口的中心就是角点。Among them, Harris corner detection is the point with sufficiently high grayscale change value in all directions within the neighborhood, and it is the point with maximum curvature on the edge curve of the image. The basic idea of corner detection is: use the corner detection operator to calculate the corner response function (Corner Response Function) for each pixel of the image, threshold the corner response function, select the threshold according to the actual situation, perform non-maximum suppression on the thresholded corner response function, and obtain non-zero points as corner points; detect corner points in the neighborhood through a small sliding window and move the window in any direction. If the grayscale value in the window changes dramatically, the center of the window is the corner point.
在一个实施例中,步骤S140中,基于所述纹理图像,采用FAST角点检测方法对所述融合后的图像进行角点检测,包括:In one embodiment, in step S140, based on the texture image, a FAST corner detection method is used to perform corner detection on the fused image, including:
在所述纹理图像上确定一点P′(x,y),在所述融合后的图像上找到与点P′(x,y)对应的点P(x,y),设定点P(x,y)的像素为I
P;
Determine a point P′(x, y) on the texture image, find a point P(x, y) corresponding to the point P′(x, y) on the fused image, and set the pixel of the point P(x, y) as IP ;
在所述融合后的图像中,以点P(x,y)为中心点,以r为半径画圆,圆上覆盖M个像素;In the fused image, a circle is drawn with point P(x, y) as the center point and r as the radius, and M pixels are covered on the circle;
设定阈值t,若点P(x,y)周围圆上的M个像素中有连续Q个点的像素小于I
P-t或者大于I
P+t,则点P(x,y)为角点;M≥Q。
Set a threshold t. If among the M pixels on the circle around point P(x,y), there are Q consecutive points with pixels less than IP -t or greater than IP +t, then point P(x,y) is a corner point; M≥Q.
具体的,在纹理图像上确定一点P′(x,y),在融合后的图像中找到点P′(x,y)对应的点P(x,y),然后对融合后的图像进行角点检测,如基于FAST角点检测方法,将点P(x,y)与该点周围圆上的16个像素点进行判断,确定点P(x,y)是否为角点,即确定点P′(x,y)是否为角点,然后依次确定纹理图像中的下个点是否为角点,直至找到纹理图像上的所有角点,结束角点检测。由于所有的角点基本都处于纹理处,通过检查纹理点是否为角点,可以缩短检测时间,提高检测效率。Specifically, a point P'(x, y) is determined on the texture image, and the point P(x, y) corresponding to the point P'(x, y) is found in the fused image. Then, the fused image is subjected to corner detection. For example, based on the FAST corner detection method, the point P(x, y) is judged with the 16 pixels on the circle around the point to determine whether the point P(x, y) is a corner point, that is, whether the point P'(x, y) is a corner point. Then, the next point in the texture image is determined to be a corner point in turn, until all the corner points on the texture image are found, and the corner detection is terminated. Since all corner points are basically located in the texture, by checking whether the texture point is a corner point, the detection time can be shortened and the detection efficiency can be improved.
示例性的,如采用FAST角点检测方法进行角点检测,FAST方法是一种基于模板和机器学习的角点检测方法,不仅计算速度快,还具有较高的精确度。FAST方法主要是考虑像素点附近的圆形窗口 上的16个像素,如下图4所示,p为中心像素点,而白框标示的点像素则是我们需要考虑的点。FAST角点定义为:若一个像素周围有一定数量的像素与该点像素值不同,则认为其为角点,具体角点检测包括以下步骤:For example, the FAST corner detection method is used for corner detection. The FAST method is a corner detection method based on templates and machine learning. It not only has fast calculation speed but also has high accuracy. The FAST method mainly considers 16 pixels on the circular window near the pixel point, as shown in Figure 4 below. p is the center pixel point, and the point pixel marked by the white box is the point we need to consider. The FAST corner point is defined as: if a certain number of pixels around a pixel have different pixel values from the point, it is considered a corner point. The specific corner point detection includes the following steps:
选取图像的一个点P,P点的像素表示为I
P。
Select a point P in the image, and the pixel at point P is denoted as I P .
以r为半径画圆,覆盖P点周围的M个像素,如下图4所示:r=3,M=16。Draw a circle with radius r to cover M pixels around point P, as shown in Figure 4 below: r=3, M=16.
设定阈值t,如果P点周围的16个像素点中有连续Q个点的像素小于I
P-t或者大于I
P+t,则该点就被认为角点。阈值t在不同场景取值有差异,若阈值t为0,则可以理解为有连续Q个点的像素大于或小于I
P的灰度值,则该点为角点。本公开实施例中,Q通常取12或9,实验表明Q取12时的角点检测性能最稳定、速度更快、效果也很好,也有实验表明Q取9时的角点检测性能更佳。具体的角点检测公式如下:
Set a threshold t. If there are Q consecutive pixels less than IP -t or greater than IP +t among the 16 pixels around point P, then the point is considered a corner point. The threshold t has different values in different scenes. If the threshold t is 0, it can be understood that there are Q consecutive pixels greater than or less than the grayscale value of IP , then the point is a corner point. In the disclosed embodiment, Q is usually 12 or 9. Experiments show that the corner point detection performance is most stable, faster, and has good results when Q is 12. There are also experiments showing that the corner point detection performance is better when Q is 9. The specific corner point detection formula is as follows:
其中,I
P表示中心点P的像素值,I
p→x表示点P的周围圆形模板x处的像素点;当点P的周围圆形模板x处的像素点I
p→x小于I
P-t,则该像素属于darker(暗些),S
p→x=d;其他两种情况分别表示brighter(亮些)和similar(相似),因此,将中心点P周围的圆形区域划分为d、s和b三种类型,统计d或b出现的次数大于Q(Q取12或9),则该点被认为是候选角点。
Wherein, IP represents the pixel value of the center point P, IP→x represents the pixel point at the circular template x around the point P; when IP→x of the pixel point at the circular template x around the point P is less than IP -t, the pixel belongs to darker, and Sp→x = d; the other two cases represent brighter and similar, respectively. Therefore, the circular area around the center point P is divided into three types: d, s and b. If the number of occurrences of d or b is greater than Q (Q is 12 or 9), the point is considered to be a candidate corner point.
在一个优选实施例中,步骤S140中,基于所述纹理图像,采用改进的FAST角点检测方法对所述融合后的图像进行角点检测,包括:In a preferred embodiment, in step S140, based on the texture image, an improved FAST corner detection method is used to perform corner detection on the fused image, including:
在所述纹理图像上确定一点P′(x,y),在所述融合后的图像上找到与点P′(x,y)对应的点P(x,y),设定点P(x,y)的像素为I
P;
Determine a point P′(x, y) on the texture image, find a point P(x, y) corresponding to the point P′(x, y) on the fused image, and set the pixel of the point P(x, y) as IP ;
在所述融合后的图像中,以点P(x,y)为中心点,以r为半径画圆,圆上覆盖M个像素;In the fused image, a circle is drawn with point P(x, y) as the center point and r as the radius, and M pixels are covered on the circle;
设定阈值t,在点P(x,y)周围圆上的M个像素中选择相互垂直方向上的四个点,若四个点中的像素值有至少三个点的像素值小于I
P-t或者大于I
P+t,则点P(x,y)为候选角点,重复判断,直至点P(x,y)周围圆上的M个像素中有连续Q个点的像素小于I
P-t或者大于I
P+t,则点P(x,y)为角点;M≥Q。
Set a threshold t, select four points in mutually perpendicular directions from the M pixels on the circle around point P(x,y), if at least three of the four points have pixel values less than IP -t or greater than IP +t, then point P(x,y) is a candidate corner point, repeat the judgment until there are Q consecutive pixels less than IP -t or greater than IP +t among the M pixels on the circle around point P(x,y), then point P(x,y) is a corner point; M≥Q.
具体的,在上述分割测试中,为了加快速度,无需对这些像素点进行逐一比较,可以先比较中心点P周围圆上水平方向和垂直方向上的四个点与中心点像素值的大小,如1、5、9、13处点的像素值,先判断这四个点中的像素值是否满足有3个或3个以上小于I
P-t或者大于I
P+t,如果不满足,则直接跳过;若果满足,则认为该点处为候选角点,并继续使用前述方法,若16个像素点中有12个点满足条件,则认为该点为角点;该角点检测方法进一步加快的角点检测速率。
Specifically, in the above segmentation test, in order to speed up the speed, there is no need to compare these pixel points one by one. The size of the pixel value of the four points in the horizontal and vertical directions on the circle around the center point P can be compared with the center point, such as the pixel values of points 1, 5, 9, and 13. First, it is determined whether the pixel values of these four points meet the condition that 3 or more are less than I P -t or greater than I P +t. If not, it is skipped directly; if it is satisfied, the point is considered to be a candidate corner point, and the above method is continued to be used. If 12 of the 16 pixel points meet the conditions, the point is considered to be a corner point; this corner point detection method further speeds up the corner point detection rate.
应该理解的是,虽然图1-图2的流程图中的各个步骤按照箭头的指示依次显示,但是这些步骤并不是必然按照箭头指示的顺序依次执行。除非本文中有明确的说明,这些步骤的执行并没有严格的顺序限制,这些步骤可以以其它的顺序执行。而且,图1-图2中的至少一部分步骤可以包括多个子步骤或者多个阶段,这些子步骤或者阶段并不必然是在同一时刻执行完成,而是可以在不同的时刻执行,这些子步骤或者阶段的执行顺序也不必然是依次进行,而是可以与其它步骤或者其它步骤的子步骤或者阶段的至少一部分轮流或者交替地执行。It should be understood that, although the various steps in the flow chart of Fig. 1-Fig. 2 are shown in sequence according to the indication of the arrows, these steps are not necessarily performed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be performed in other orders. Moreover, at least a portion of the steps in Fig. 1-Fig. 2 may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.
应当注意,尽管在附图中以特定顺序描述了本公开方法的操作,但是,这并非要求或者暗示必须按照该特定顺序来执行这些操作,或是必须执行全部所示的操作才能实现期望的结果。相反,流程图中描绘的步骤可以改变执行顺序。It should be noted that although the operations of the disclosed method are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in this specific order, or that all the operations shown must be performed to achieve the desired results. On the contrary, the steps depicted in the flowchart can change the execution order.
参考图5所示的基于纹理特征的角点检测***的示例性结构框图,本实施例的一种基于纹理特征的角点检测***200,包括:Referring to the exemplary structural block diagram of a corner point detection system based on texture features shown in FIG5 , a corner point detection system 200 based on texture features in this embodiment includes:
纹理提取单元220,配置成对待处理图像进行纹理特征提取,得到纹理图像;The texture extraction unit 220 is configured to extract texture features from the image to be processed to obtain a texture image;
图像融合单元230,配置成将所述纹理图像与所述待处理图像进行融合,得到融合后的图像;An image fusion unit 230, configured to fuse the texture image with the image to be processed to obtain a fused image;
角点检测单元240,配置成基于所述纹理图像,对所述融合后的图像进行角点检测。The corner point detection unit 240 is configured to perform corner point detection on the fused image based on the texture image.
具体的,本公开实施例提供的***,通过将待处理图像的背景与纹理进行分离,得到纹理图像,基于图像的纹理特征进行角点检测,确定纹理特征是否为角点,所选择的像素位置能更好的反应角点的性能,可以大大节省算力和耗时。Specifically, the system provided by the embodiments of the present disclosure separates the background and texture of the image to be processed to obtain a texture image, performs corner point detection based on the texture features of the image, and determines whether the texture features are corner points. The selected pixel positions can better reflect the performance of the corner points, which can greatly save computing power and time.
在一个实施例中,如图6所示,所述***还包括:In one embodiment, as shown in FIG6 , the system further includes:
图像滤波单元210,所述图像滤波单元配置成对所述待处理图像进行滤波处理,得到滤波后的图像。The image filtering unit 210 is configured to perform filtering on the image to be processed to obtain a filtered image.
具体的,通过图像滤波单元210对待处理图像进行滤波处理可以去除图像中的噪声,避免将图像中的噪声点被误认为角点,提高角点检测的准确性。图像滤波单元210可配置成执行高斯滤波法、Gabor滤波法。Specifically, the image filtering unit 210 can filter the image to be processed to remove noise in the image, avoid mistaking noise points in the image for corner points, and improve the accuracy of corner point detection. The image filtering unit 210 can be configured to perform Gaussian filtering or Gabor filtering.
在一个实施例中,所述图像滤波单元210,还配置成采用不同尺度因子、不同的半径的高斯核函数对待处理图像进行尺度处理,得到滤波后的图像。In one embodiment, the image filtering unit 210 is further configured to use Gaussian kernel functions with different scale factors and different radii to perform scale processing on the image to be processed to obtain a filtered image.
具体的,图像滤波单元210采用高斯滤波的方法对待处理图像进行滤波处理,具体的高斯滤波方法如上所述,本公开实施例不再一一赘述。Specifically, the image filtering unit 210 uses a Gaussian filtering method to filter the image to be processed. The specific Gaussian filtering method is as described above, and the embodiments of the present disclosure will not be described in detail.
在一个实施例中,所述纹理提取单元220,配置成根据待处理图像以及不同尺度的滤波后的图像得到多个细节图像,并为多个细节图像分别配置相应的权重以获取纹理图像。In one embodiment, the texture extraction unit 220 is configured to obtain a plurality of detail images according to the image to be processed and filtered images of different scales, and configure corresponding weights for the plurality of detail images respectively to obtain a texture image.
具体的,纹理提取单元220配置成对待处理图像以及不同尺度的滤波后的图像进行处理,获取纹理图像,具体纹理图像的提取方法如上所述。Specifically, the texture extraction unit 220 is configured to process the image to be processed and the filtered images of different scales to obtain a texture image. The specific method for extracting the texture image is as described above.
在一个实施例中,所述图像融合单元230,配置成将所述纹理图像与所述待处理图像进行相加,得到融合后的图像。In one embodiment, the image fusion unit 230 is configured to add the texture image and the image to be processed to obtain a fused image.
在一个实施例中,所述角点检测单元240,配置成执行如下至少 一种方法:SIFT角点检测、ORB角点检测、FAST角点检测、Harris角点检测、SURF角点检测。In one embodiment, the corner detection unit 240 is configured to perform at least one of the following methods: SIFT corner detection, ORB corner detection, FAST corner detection, Harris corner detection, and SURF corner detection.
具体的,角点检测单元240执行的检测方法如上所述,本公开实施例不再一一赘述。Specifically, the detection method performed by the corner point detection unit 240 is as described above, and the embodiments of the present disclosure will not be described in detail.
本公开实施例的第三方面,提供了一种电子设备,其内部结构图可以如图7所示。该电子设备包括通过***总线连接的一个或多个处理器、存储器、通信接口、显示屏和输入装置。其中,该电子设备的一个或多个处理器配置成提供计算和控制能力。该电子设备的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作***和计算机可读指令。该内存储器为非易失性存储介质中的操作***和计算机可读指令的运行提供环境。该电子设备的通信接口配置成与外部的终端进行有线或无线方式的通信,无线方式可通过WIFI、运营商网络、近场通信(NFC)或其他技术实现。该计算机可读指令被一个或多个处理器执行时以实现一种基于纹理特征的角点检测方法。该电子设备的显示屏可以是液晶显示屏或者电子墨水显示屏,该电子设备的输入装置可以是显示屏上覆盖的触摸层,也可以是电子设备外壳上设置的按键、轨迹球或触控板,还可以是外接的键盘、触控板或鼠标等。In a third aspect of the embodiments of the present disclosure, an electronic device is provided, and its internal structure diagram may be shown in FIG7. The electronic device includes one or more processors, memory, communication interface, display screen and input device connected via a system bus. Among them, the one or more processors of the electronic device are configured to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer-readable instructions. The internal memory provides an environment for the operation of the operating system and computer-readable instructions in the non-volatile storage medium. The communication interface of the electronic device is configured to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be implemented through WIFI, an operator network, near field communication (NFC) or other technologies. When the computer-readable instructions are executed by one or more processors, a corner detection method based on texture features is implemented. The display screen of the electronic device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the electronic device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the electronic device, or an external keyboard, touchpad or mouse, etc.
本领域技术人员可以理解,图7中示出的结构,仅仅是与本公开方案相关的部分结构的框图,并不构成对本公开方案所应用于其上的电子设备的限定,具体的电子设备可以包括比图中所示更多或更少的部件,或者组合某些部件,或者具有不同的部件布置。Those skilled in the art will understand that the structure shown in FIG. 7 is merely a block diagram of a partial structure related to the scheme of the present disclosure, and does not constitute a limitation on the electronic device to which the scheme of the present disclosure is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.
在一个实施例中,本公开提供的基于纹理特征的角点检测***200可以实现为一种计算机可读指令的形式,计算机可读指令可在如图7所示的电子设备上运行。电子设备的存储器中可存储组成该基于纹理特征的角点检测***200的各个程序模块,比如,图6所示的图像滤波单元210、纹理提取单元220、图像融合单元230、角点检测单元240等。各个程序模块构成的计算机可读指令使得一个或多个处理器执行本说明书中描述的本公开各个实施例的基于纹理特征的角点检测方法中的步骤。In one embodiment, the texture feature-based corner detection system 200 provided by the present disclosure may be implemented in the form of a computer-readable instruction, and the computer-readable instruction may be run on an electronic device as shown in FIG7. The memory of the electronic device may store various program modules constituting the texture feature-based corner detection system 200, such as the image filtering unit 210, the texture extraction unit 220, the image fusion unit 230, the corner detection unit 240, etc. shown in FIG6. The computer-readable instructions constituted by the various program modules enable one or more processors to execute the steps of the texture feature-based corner detection method of various embodiments of the present disclosure described in this specification.
在一个实施例中,提供了一种电子设备,包括存储器和一个或多个处理器,该存储器存储有计算机可读指令,该一个或多个处理器执行计算机可读指令时实现以下上述基于纹理特征的角点检测方法的步骤。In one embodiment, an electronic device is provided, including a memory and one or more processors, wherein the memory stores computer-readable instructions, and the one or more processors implement the following steps of the above-mentioned texture feature-based corner detection method when executing the computer-readable instructions.
在一个实施例中,还提供了一个或多个存储有计算机可读指令的非易失性计算机可读存储介质,其上存储有计算机可读指令,计算机可读指令被一个或多个处理器执行时实现上述各方法实施例中的步骤。In one embodiment, one or more non-volatile computer-readable storage media storing computer-readable instructions are also provided, on which computer-readable instructions are stored. When the computer-readable instructions are executed by one or more processors, the steps in the above-mentioned method embodiments are implemented.
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程,是可以通过计算机可读指令来指令相关的硬件来完成,所述的计算机可读指令可存储于一非易失性计算机可读取存储介质中,该计算机可读指令在执行时,可包括如上述各方法的实施例的流程。其中,本公开所提供的各实施例中所使用的对存储器、数据库或其它介质的任何引用,均可包括非易失性和易失性存储器中的至少一种。非易失性存储器可包括只读存储器(Read-Only Memory,ROM)、磁带、软盘、闪存或光存储器等。易失性存储器可包括随机存取存储器(Random Access Memory,RAM)或者外部高速缓冲存储器。作为说明而非局限,RAM以多种形式可得,比如静态随机存取存储器(Static Random Access Memory,SRAM)和动态随机存取存储器(Dynamic Random Access Memory,DRAM)等。Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a non-volatile computer-readable storage medium. When the computer-readable instructions are executed, they can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided by the present disclosure can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static random access memory (SRAM) and dynamic random access memory (DRAM).
以上实施例的各技术特征可以进行任意的组合,为使描述简洁,未对上述实施例中的各个技术特征所有可能的组合都进行描述,然而,只要这些技术特征的组合不存在矛盾,都应当认为是本说明书记载的范围。The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
以上所述实施例仅表达了本公开的几种实施方式,其描述较为具体和详细,但并不能因此而理解为对发明专利范围的限制。应当指出的是,对于本领域的普通技术人员来说,在不脱离本公开构思的前提下,还可以做出若干变形和改进,这些都属于本公开的保护范围。因此,本公开专利的保护范围应以所附权利要求为准。The above-mentioned embodiments only express several implementation methods of the present disclosure, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present disclosure, and these all belong to the protection scope of the present disclosure. Therefore, the protection scope of the patent of the present disclosure shall be subject to the attached claims.
本公开提供的基于纹理特征的角点检测方法,角点的检测不依赖于检测点的顺序和角点附近的分布,由于角点基本位于纹理特征处,因此,基于图像的纹理特征进行角点检测,所选择的像素位置能更好的反应角点的性能,检测时间短,检测效率和准确性高,具有很强的工业实用性。The present invention provides a corner point detection method based on texture features. The detection of corner points does not depend on the order of detection points and the distribution near the corner points. Since the corner points are basically located at the texture features, the corner point detection is performed based on the texture features of the image. The selected pixel positions can better reflect the performance of the corner points. The detection time is short, the detection efficiency and accuracy are high, and the method has strong industrial applicability.
Claims (15)
- 一种基于纹理特征的角点检测方法,其特征在于,包括:A corner point detection method based on texture features, characterized by comprising:对待处理图像进行纹理特征提取,得到纹理图像;Extract texture features of the image to be processed to obtain a texture image;将所述纹理图像与所述待处理图像进行融合,得到融合后的图像;Fusing the texture image with the image to be processed to obtain a fused image;基于所述纹理图像,对所述融合后的图像进行角点检测。Based on the texture image, corner point detection is performed on the fused image.
- 根据权利要求1所述的基于纹理特征的角点检测方法,其中,所述对待处理图像进行纹理特征提取之前还包括:The method for detecting corner points based on texture features according to claim 1, wherein before extracting texture features from the image to be processed, the method further comprises:对所述待处理图像进行滤波处理,得到滤波后的图像。Perform filtering on the image to be processed to obtain a filtered image.
- 根据权利要求2所述的基于纹理特征的角点检测方法,其中,所述滤波处理的方法包括至少如下一种:高斯滤波法、Gabor滤波法。According to the texture feature-based corner detection method of claim 2, wherein the filtering processing method includes at least one of the following: Gaussian filtering method and Gabor filtering method.
- 根据权利要求3所述的基于纹理特征的角点检测方法,其中,采用高斯滤波法对所述待处理图像进行滤波处理包括:According to the texture feature-based corner detection method of claim 3, wherein the filtering process of the image to be processed by using the Gaussian filtering method comprises:采用不同尺度因子、不同的半径的高斯核函数对待处理图像进行尺度处理,得到滤波后的图像:Use Gaussian kernel functions with different scale factors and different radii to scale the image to be processed and obtain the filtered image:I(x,y) Ri=Gaussian i(I(x,y) in);i=[1,2,…,N] I(x,y) Ri = Gaussian i (I(x,y) in ); i = [1,2,…,N]其中,i表示滤波次数,Gaussian i表示第i个尺度因子为σ i、半径为R i的高斯核函数,I(x,y) in表示待处理图像,I(x,y) Ri表示第i个滤波后的图像。 Where i represents the number of filtering times, Gaussian i represents the Gaussian kernel function with the i-th scale factor σ i and radius R i , I(x,y) in represents the image to be processed, and I(x,y) Ri represents the i-th filtered image.
- 根据权利要求4所述的基于纹理特征的角点检测方法,其中,对所述滤波后的图像进行纹理特征提取,得到纹理图像,公式如下:According to the corner detection method based on texture features of claim 4, wherein texture features are extracted from the filtered image to obtain a texture image, and the formula is as follows:其中,I(x,y) Detaili表示第i个细节图像;I(x,y) Detail表示纹理图像,w i 表示第i个权重值。 Among them, I(x,y) Detaili represents the i-th detail image; I(x,y) Detail represents the texture image, and wi represents the i-th weight value.
- 根据权利要求5所述的基于纹理特征的角点检测方法,其中,所述融合后的图像I(x,y) out为: The method for corner detection based on texture features according to claim 5, wherein the fused image I(x, y) out is:I(x,y) out=I(x,y) in+I(x,y) Detail。 I(x,y) out =I(x,y) in +I(x,y) Detail .
- 根据权利要求1-6任一项所述的基于纹理特征的角点检测方法,其中,所述角点检测的方法包括至少如下一种:SIFT角点检测、ORB角点检测、FAST角点检测、Harris角点检测、SURF角点检测。According to the texture feature-based corner detection method according to any one of claims 1 to 6, wherein the corner detection method comprises at least one of the following: SIFT corner detection, ORB corner detection, FAST corner detection, Harris corner detection, and SURF corner detection.
- 一种基于纹理特征的角点检测***,其特征在于,包括:A corner point detection system based on texture features, characterized by comprising:纹理提取单元,配置成对待处理图像进行纹理特征提取,得到纹理图像;A texture extraction unit configured to extract texture features from the image to be processed to obtain a texture image;图像融合单元,配置成将所述纹理图像与所述待处理图像进行融合,得到融合后的图像;an image fusion unit, configured to fuse the texture image with the image to be processed to obtain a fused image;角点检测单元,配置成基于所述纹理图像,对所述融合后的图像进行角点检测。The corner point detection unit is configured to perform corner point detection on the fused image based on the texture image.
- 根据权利要求8所述的角点检测***,其中,所述***还包括:The corner detection system according to claim 8, wherein the system further comprises:图像滤波单元,配置成对所述待处理图像进行滤波处理,得到滤波后的图像。The image filtering unit is configured to perform filtering on the image to be processed to obtain a filtered image.
- 根据权利要求9所述的角点检测***,其中,所述图像滤波单元,还配置成采用不同尺度因子、不同的半径的高斯核函数对待处理图像进行尺度处理,得到滤波后的图像:The corner detection system according to claim 9, wherein the image filtering unit is further configured to use Gaussian kernel functions with different scale factors and different radii to scale the image to be processed to obtain a filtered image:I(x,y) Ri=Gaussian i(I(x,y) in);i=[1,2,…,N] I(x,y) Ri = Gaussian i (I(x,y) in ); i = [1,2,…,N]其中,i表示滤波次数,Gaussian i表示第i个尺度因子为σ i、半径为R i的高斯核函数,I(x,y) in表示待处理图像,I(x,y) Ri表示第i个滤波后的图像。 Where i represents the number of filtering times, Gaussian i represents the Gaussian kernel function with the i-th scale factor σ i and radius R i , I(x,y) in represents the image to be processed, and I(x,y) Ri represents the i-th filtered image.
- 根据权利要求10所述的角点检测***,其中,所述纹理提取单元,配置成对所述滤波后的图像进行纹理特征提取,得到纹理图像,公式如下:The corner detection system according to claim 10, wherein the texture extraction unit is configured to extract texture features from the filtered image to obtain a texture image, and the formula is as follows:其中,I(x,y) Detaili表示第i个细节图像;I(x,y) Detail表示纹理图像,w i表示第i个权重值。 Among them, I(x,y) Detaili represents the i-th detail image; I(x,y) Detail represents the texture image, and wi represents the i-th weight value.
- 根据权利要求11所述的角点检测***,其中,所述图像融合单元,配置成将所述纹理图像与所述待处理图像进行相加,得到融合后的图像;The corner detection system according to claim 11, wherein the image fusion unit is configured to add the texture image to the image to be processed to obtain a fused image;所述融合后的图像I(x,y) out为: The fused image I(x,y) out is:I(x,y) out=I(x,y) in+I(x,y) Detail。 I(x,y) out =I(x,y) in +I(x,y) Detail .
- 根据权利要求8-13任一项所述的角点检测***,其中,所述角点检测单元,配置成执行如下至少一种方法:SIFT角点检测、ORB角点检测、FAST角点检测、Harris角点检测、SURF角点检测。The corner detection system according to any one of claims 8 to 13, wherein the corner detection unit is configured to perform at least one of the following methods: SIFT corner detection, ORB corner detection, FAST corner detection, Harris corner detection, and SURF corner detection.
- 一种电子设备,包括存储器和一个或多个处理器,所述存储器存储有计算机可读指令,其特征在于,所述一个或多个处理器执行所述计算机可读指令时实现权利要求1-7中任一项所述的基于纹理特征的角点检测方法的步骤。An electronic device comprises a memory and one or more processors, wherein the memory stores computer-readable instructions, and wherein the one or more processors implement the steps of the texture feature-based corner detection method described in any one of claims 1 to 7 when executing the computer-readable instructions.
- 一个或多个存储有计算机可读指令的非易失性计算机可读存储介质,其上存储有计算机可读指令,其特征在于,所述计算机可读指令被一个或多个处理器执行时实现权利要求1-7中任一项所述的基于纹理特征的角点检测方法的步骤。One or more non-volatile computer-readable storage media storing computer-readable instructions, on which computer-readable instructions are stored, characterized in that when the computer-readable instructions are executed by one or more processors, the steps of the texture feature-based corner detection method described in any one of claims 1-7 are implemented.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211178236.5A CN115661036A (en) | 2022-09-26 | 2022-09-26 | Corner point detection method and system based on texture features, electronic device and medium |
CN202211178236.5 | 2022-09-26 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2024066090A1 true WO2024066090A1 (en) | 2024-04-04 |
Family
ID=84985341
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2022/141469 WO2024066090A1 (en) | 2022-09-26 | 2022-12-23 | Corner detection method and system based on texture features, electronic device, and medium |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN115661036A (en) |
WO (1) | WO2024066090A1 (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105913463A (en) * | 2016-04-11 | 2016-08-31 | 中南大学 | Position prior principle-based texture-color characteristic overall saliency detection method |
AU2020100199A4 (en) * | 2020-02-08 | 2020-03-19 | Cao, Sihua MR | A medical image fusion method based on two-layer decomposition and improved spatial frequency |
US20210174489A1 (en) * | 2019-12-04 | 2021-06-10 | Beijing Boe Optoelectronics Technology Co., Ltd. | Method and apparatus for detecting a screen, and electronic device |
WO2021249233A1 (en) * | 2020-06-10 | 2021-12-16 | 中铁四局集团有限公司 | Image processing method, target recognition model training method, and target recognition method |
WO2022095742A1 (en) * | 2020-11-05 | 2022-05-12 | 晶晨半导体(上海)股份有限公司 | Image and video processing methods and systems, and data processing device and medium |
CN114998162A (en) * | 2022-06-23 | 2022-09-02 | 上海闻泰信息技术有限公司 | Moving object detection method and device, terminal equipment and storage medium |
-
2022
- 2022-09-26 CN CN202211178236.5A patent/CN115661036A/en active Pending
- 2022-12-23 WO PCT/CN2022/141469 patent/WO2024066090A1/en unknown
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105913463A (en) * | 2016-04-11 | 2016-08-31 | 中南大学 | Position prior principle-based texture-color characteristic overall saliency detection method |
US20210174489A1 (en) * | 2019-12-04 | 2021-06-10 | Beijing Boe Optoelectronics Technology Co., Ltd. | Method and apparatus for detecting a screen, and electronic device |
AU2020100199A4 (en) * | 2020-02-08 | 2020-03-19 | Cao, Sihua MR | A medical image fusion method based on two-layer decomposition and improved spatial frequency |
WO2021249233A1 (en) * | 2020-06-10 | 2021-12-16 | 中铁四局集团有限公司 | Image processing method, target recognition model training method, and target recognition method |
WO2022095742A1 (en) * | 2020-11-05 | 2022-05-12 | 晶晨半导体(上海)股份有限公司 | Image and video processing methods and systems, and data processing device and medium |
CN114998162A (en) * | 2022-06-23 | 2022-09-02 | 上海闻泰信息技术有限公司 | Moving object detection method and device, terminal equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN115661036A (en) | 2023-01-31 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10896349B2 (en) | Text detection method and apparatus, and storage medium | |
US10936911B2 (en) | Logo detection | |
CN110414507B (en) | License plate recognition method and device, computer equipment and storage medium | |
US20200005022A1 (en) | Method, terminal, and storage medium for tracking facial critical area | |
CN110852349A (en) | Image processing method, detection method, related equipment and storage medium | |
CN104751093B (en) | Method and apparatus for obtaining the video identification code that host equipment is shown | |
CN111640089A (en) | Defect detection method and device based on feature map center point | |
CN111292377B (en) | Target detection method, device, computer equipment and storage medium | |
CN114155365B (en) | Model training method, image processing method and related device | |
CN112101386B (en) | Text detection method, device, computer equipment and storage medium | |
CN111814905A (en) | Target detection method, target detection device, computer equipment and storage medium | |
CN110807110B (en) | Image searching method and device combining local and global features and electronic equipment | |
CN113298809B (en) | Composite material ultrasonic image defect detection method based on deep learning and superpixel segmentation | |
WO2022194079A1 (en) | Sky region segmentation method and apparatus, computer device, and storage medium | |
CN114444565A (en) | Image tampering detection method, terminal device and storage medium | |
CN113449538A (en) | Visual model training method, device, equipment and storage medium | |
WO2018053710A1 (en) | Morphological processing method of digital image and digital image processing device | |
CN114494751A (en) | License information identification method, device, equipment and medium | |
WO2024066090A1 (en) | Corner detection method and system based on texture features, electronic device, and medium | |
CN113723375B (en) | Double-frame face tracking method and system based on feature extraction | |
CN111753722B (en) | Fingerprint identification method and device based on feature point type | |
CN112465931B (en) | Image text erasing method, related equipment and readable storage medium | |
CN111402168B (en) | Image target correction method and device, terminal and storage medium | |
CN115205555B (en) | Method for determining similar images, training method, information determining method and equipment | |
CN113095211B (en) | Image processing method, system and electronic equipment |
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: 22960685 Country of ref document: EP Kind code of ref document: A1 |