Image edge enhancement methods, apparatus, devices and storage media

Abstract

This invention discloses an image edge enhancement method, apparatus, device, and storage medium, belonging to the field of image processing. The method includes: acquiring the gravity gradient signal of the original image and calculating standardized gravity gradient tensor data; calculating the horizontal, vertical, and depth gradient magnitudes and the total horizontal gradient magnitude based on the tensor data; then extracting ridge features, tilt angle zero-value line features, and adaptive Canny edge features respectively, and fusing the three to obtain an initial edge contour; further performing morphological processing on the initial edge contour to obtain an enhanced edge contour; finally, smoothing the enhanced edge contour and embedding it into the original image to output the edge-enhanced image. This invention, by fusing three contour features, can enhance the edge response of weak and small targets, effectively reducing the false alarm rate.

Description

Technical Field

[0001] This invention relates to the field of image processing, and more particularly to an image edge enhancement method, apparatus, device, and storage medium. Background Technology

[0002] With the development of computer vision technologies, the application scenarios for object detection and segmentation are becoming increasingly widespread. When detecting object edges in images, complex and variable lighting conditions can lead to uneven brightness, overexposure, or underexposure in local image regions. Existing techniques based on local gradients or single-modal features lack an understanding of the global semantics of the image, are susceptible to noise, and suffer from blurred or broken edges, resulting in low detection accuracy. Furthermore, while some methods based on multi-source data fusion can improve edge integrity, these methods have high data acquisition costs and lack in-depth analysis of the physical properties of gravity gradient data.

[0003] As can be seen from the above, traditional methods for applying gravity gradient signals mainly focus on obtaining gradient field results and fusing multi-source data. However, these methods still have room for improvement in addressing the problem of high false alarm rates at image target edges and the inability to accurately depict the contour intensity of small or even "weak" targets. Summary of the Invention

[0004] This invention provides an image edge enhancement method, apparatus, device, and storage medium, aiming to solve the problems of existing edge detection methods based on local gradients being susceptible to noise, having insufficient response to weak and small target edges, and having a high false alarm rate.

[0005] In a first aspect, embodiments of the present invention provide an image edge enhancement method, comprising: The gravity gradient signal of the original image is obtained, and the normalized gravity gradient tensor data is obtained by calculating and preprocessing the gravity gradient signal. The normalized gravity gradient tensor data is used to calculate the gradient field amplitude, which includes: the gradient field amplitude in the horizontal direction, the gradient field amplitude in the vertical direction, the gradient field amplitude in the depth direction, and the total horizontal gradient field amplitude. Calculating and extracting multi-source contour features from the gradient field amplitude, wherein the calculation and extraction of multi-source contour features from the gradient field amplitude includes: Extract ridge features, which depict edge contours from the perspective of a linear geometric skeleton; Extract the zero-value line feature of the tilt angle, and use the boundary zero intersection point to delineate the edge contour; Extract adaptive Canny edge features, which generate edge contours using local contrast abrupt changes; The initial edge contour is obtained by fusing the multi-source contour features; Morphological processing is performed on the initial edge contour to obtain an enhanced edge contour; The enhanced edge contour is smoothed and then embedded into the original image to obtain an image with enhanced target edges.

[0006] Secondly, embodiments of the present invention also provide an image edge enhancement device, comprising: The data acquisition module is used to acquire the gravity gradient signal of the original image, and to obtain standardized gravity gradient tensor data by calculating and preprocessing the gravity gradient signal. The gradient calculation module is used to calculate the horizontal gradient field amplitude, the vertical gradient field amplitude, the depth gradient field amplitude, and the total horizontal gradient field amplitude based on the standardized gravity gradient tensor data. The feature extraction module is used to extract multi-source contour features from the gradient field amplitude: ridge features, which depict the edge contour from the perspective of a linear geometric skeleton; tilt angle zero-value line features, which delineate the edge contour using boundary zero intersection points; and adaptive Canny edge features, which generate the edge contour using local contrast abrupt changes. The fusion module is used to fuse the multi-source contour features to obtain an initial edge contour; The morphological processing module is used to perform morphological processing on the initial edge contour to obtain an enhanced edge contour. The output module is used to embed the enhanced edge contour into the original image to obtain an image with enhanced target edges.

[0007] Thirdly, embodiments of the present invention also provide an apparatus, including one or more processors and a storage device, wherein the storage device stores one or more programs, and when the programs are executed by the processor, the above-described image edge enhancement method is implemented.

[0008] Fourthly, embodiments of the present invention also provide a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform the image edge enhancement method provided in the above embodiments.

[0009] The image edge enhancement method, apparatus, device, and storage medium provided in this invention acquire the gravity gradient signal of the original image and calculate the standardized gravity gradient tensor data. Based on the data, the gradient field amplitude in each direction is calculated. Then, ridge features, tilt angle zero-value line features, and adaptive Canny edge features are extracted. These three features are added pixel-level to obtain an initial edge contour. Morphological processing is then used to smooth the contour to obtain the target edge contour. Finally, the enhanced image is embedded into the original image. Specifically, ridge features depict the edge contour from the perspective of a linear geometric skeleton, resulting in lines with strong continuity; tilt angle zero-value line features use boundary zero-intersection points to delineate the edge contour, achieving a high capture rate for weak signals; and adaptive Canny edge features generate the edge contour using local contrast abrupt changes, resulting in a low false alarm rate for edge recognition. The initial edge contour obtained by fusing these three contour features possesses the advantages of all three, effectively improving the edge response of weak targets and small anomalies while reducing the false alarm rate. Attached Figure Description

[0010] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a schematic flowchart of the image edge enhancement method provided in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the image edge enhancement device provided in Embodiment 2 of the present invention; Figure 3 This is a schematic diagram of the device provided in Embodiment 3 of the present invention. Detailed Implementation

[0011] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention, and not all of the structures.

[0012] Example 1 Figure 1 This is a flowchart illustrating the image edge enhancement method provided in Embodiment 1 of the present invention. Using this method, the original gravity gradient data can be transformed into a target edge-enhanced contour, which can then be embedded into the original image for subsequent use. This method can be executed by an image edge enhancement device and specifically includes the following steps.

[0013] Step 110: Obtain the gravity gradient signal of the original image, and obtain standardized gravity gradient tensor data by calculating and preprocessing the gravity gradient signal.

[0014] Get the original imageI raw For example, images that traditional image edge recognition methods cannot accurately process due to noise interference are used as the original images, such as images with broken or incomplete edges, images with uneven brightness, and overexposed images. Original Image I raw The resolution size is u v, the pixels in the image are arranged in spatial order to form the original image matrix as the original image. I raw The gravity gradient signal. Where the matrix contains... Corresponding image I raw The pixel in row u and column v of the image. I raw As shown below: I raw= ; After obtaining the gravity gradient signal of the original image, the gravity potential is extracted from the gravity gradient signal for subsequent processing. For example, a forward Fourier transform is performed on the gravity gradient signal of the original image. The result of the forward Fourier transform is divided by the Laplace operator expression to obtain the quotient. An inverse Fourier transform is then performed on the quotient to obtain the gravity potential G. This can be calculated as follows: G= ; in, (·) represents the inverse Fourier transform. F (·) represents the positive Fourier transform. Here is the expression for the Laplace operator, and G is the gravitational potential.

[0015] After calculating the gravity potential G in the original gravity gradient signal, the matrix obtained by taking the second-order partial derivative of the gravity potential G is used as the two-dimensional gravity gradient data. Preferably, the matrix can be a 3×3 matrix. For example, if the rows and columns of the matrix are arranged in the order of x, y, z, then the matrix contains 9 gravity gradient tensor data points. , , , , , , , and It can be calculated in the following way: ; in This is two-dimensional gravity gradient data. This represents the second-order partial derivative of G obtained by taking the partial derivative of G twice with respect to x. This represents the second-order partial derivative of G obtained by taking the partial derivatives of G with respect to x and y respectively. This represents the second-order partial derivative of G obtained by taking the partial derivatives of G with respect to x and z respectively. This represents the second-order partial derivatives of G obtained by taking the partial derivatives of G with respect to y and x respectively. This represents the second-order partial derivative obtained by taking the partial derivative of G with respect to y twice. This represents the second-order partial derivatives of G obtained by taking the partial derivatives of G with respect to y and z respectively. This represents the second-order partial derivatives of G obtained by taking the partial derivatives of G with respect to z and x respectively. This represents the sum of the second-order partial derivatives of G obtained by taking the partial derivatives of G with respect to z and y respectively. Let G be the second-order partial derivative obtained by taking two partial derivatives of G with respect to z.

[0016] Because the original image contains missing or broken edges, uneven brightness, or overexposure, the resulting two-dimensional gravity gradient data contains noise, so noise reduction processing is required.

[0017] This embodiment designs a smoothing filter for noise reduction. Optionally, a smoothing filter H(w,w) of size w×w is set. The choice of filter size w needs to strike a balance between noise reduction effect and edge preservation: the larger w is, the stronger the noise reduction capability, but it may blur the real edges; the smaller w is, the better the edge preservation, but the noise reduction is insufficient. It can be calculated as follows: H(w,w)= ; in For the smoothing filter line, number The column's weight value, and , Satisfying 1≤ ≤w,1≤ ≤w ， H(w,w) is a smoothing filter.

[0018] The obtained two-dimensional gravity gradient data is smoothed and denoised using a filter. For each pixel (i,j) in the image, the weighted average of all pixels in its neighborhood is taken as the filtering result for that point, resulting in the denoised two-dimensional gravity gradient data G. nonoise Represented as a matrix, preferably a 3×3 matrix, containing 、 、 、 、 、 、 、 and These are the nine components of the denoised gravity gradient tensor data. They can be calculated as follows: G nonoise = = ; in, N This represents the total number of pixels within the window. Therefore ( p , q The neighborhood centered on ) for Two-dimensional gravity gradient data in coordinate system; The data is the denoised gravity gradient tensor. m , n ∈( x,y,z ).

[0019] After denoising the two-dimensional gravity gradient data, the resulting nine components differ in physical meaning and measurement range. If these unprocessed components are used together, the larger numerical values ​​will dominate subsequent gradient field amplitude calculations and feature extraction, masking weak signals contained in the smaller numerical values ​​and thus reducing the sensitivity and accuracy of edge detection. Optionally, before feature extraction, the components can be standardized to eliminate differences in numerical ranges, ensuring that each component has the same weight in subsequent calculations.

[0020] For example, the nine components of the denoised two-dimensional gradient gravity data are summed, and the sum is divided by the total number of pixels in the image to obtain the mean. The standard deviation is then calculated using the mean. The mean is subtracted from the denoised two-dimensional gradient tensor data, and the result is divided by the standard deviation to obtain the standardized gravity gradient tensor data G. score ,Include 、 、 、 、 、 、 、 and The components of these nine standardized gravity gradient tensors can be calculated as follows: ; ; Where u v represents the total number of pixels in the image, and m, n∈(x,y,z) represent the nine components of the denoised gravity gradient tensor. , These are the mean and standard deviation of the denoised two-dimensional gradient gravity data, respectively.

[0021] G score = = ; in, , G represents the mean and standard deviation of the denoised two-dimensional gradient gravity data, respectively. score To standardize the gravity gradient tensor data.

[0022] Step 120: Calculate the normalized gravity gradient tensor data to obtain the gradient field amplitude, which includes: horizontal gradient field amplitude, vertical gradient field amplitude, depth gradient field amplitude, and total horizontal gradient field amplitude.

[0023] The magnitudes of the four gradient fields mentioned above are mathematically referred to as horizontal fields. x Orientation gradient field amplitude, horizontal y Orientation gradient field amplitude, horizontal z The magnitude of the directional gradient field and the magnitude of the total horizontal gradient field.

[0024] For example, by analyzing the components of the standardized gravity gradient tensor data in the three horizontal x-directions and 、 and Squaring each value individually and then summing them, then taking the square root of the sum gives the corresponding gradient field magnitude in the horizontal x-direction. It can be calculated in the following way: = ; For example, by analyzing the components of the standardized gravity gradient tensor data in the three horizontal y-directions and 、 and Squaring each value individually and then summing them, then taking the square root of the sum gives the corresponding gradient field magnitude in the horizontal y-direction. It can be calculated in the following way: = ; For example, by analyzing the components of the standardized gravity gradient tensor data in the three horizontal z-directions and 、 and Squaring each value individually and then summing them, then taking the square root of the sum gives the corresponding gradient field magnitude in the horizontal z-direction. It can be calculated in the following way: = ; For example, the gradient field magnitude in the horizontal x-direction obtained above Horizontal y-direction gradient field amplitude and the magnitude of the gradient field in the horizontal z-direction The total horizontal gradient field magnitude is obtained by squaring each element separately, summing the results, and then taking the square root of the summation. thg It can be calculated in the following way: thg= ; The gradient field magnitude obtained above can be used for subsequent extraction of contour features.

[0025] Step 130: Calculate and extract multi-source contour features from the gradient field amplitude. This calculation and extraction of multi-source contour features from the gradient field amplitude includes: Extract ridge features, which depict edge contours from the perspective of a linear geometric skeleton; Extract the zero-value line feature of the tilt angle, and use the boundary zero intersection point to delineate the edge contour; An adaptive Canny edge feature is extracted, which generates an edge contour using local contrast abrupt changes.

[0026] This embodiment is based on the obtained standardized gravity gradient tensor data and the gradient field amplitude in the horizontal x-direction. Horizontal y-direction gradient field amplitude Horizontal z-direction gradient field magnitude and the total horizontal gradient field amplitude thg Three types of contour features are extracted: ridge features, tilt angle zero-value line features, and adaptive Canny edge features. These three features respectively capture the linearly constructed geometric skeleton, the zero-intersection points at the boundary locations, and the local contrast abrupt changes.

[0027] An effective method for ridge feature extraction is eigenvalue analysis using the Hessian matrix, where the Hessian matrix is ​​a square matrix composed of all second-order partial derivatives of a multivariate function. Optionally, the normalized gravity gradient tensor data G obtained in step 110 can be used... scoreAs a Hessian matrix, eigenvalue decomposition is performed on the Hessian matrix. Points where the second eigenvalue is negative and the absolute value of the first eigenvalue is greater than the absolute value of the largest first eigenvalue are taken as ridge points, generating a ridge intensity map as the ridge feature. This can be calculated as follows: =(λ2<0) and (|λ1|>threshold T_ridge); Where the threshold T_ridge is the percentage of the maximum |λ1|, and λ1, λ2 are G score The eigenvalues ​​of the matrix, satisfying |λ1|≥|λ2|. This is the ridge line point.

[0028] Fea1= ; Fea1 is a ridge feature. The ridge point is a feature of the ridge line.

[0029] Ridge features depict edge contours from the perspective of linear geometric skeletons. Compared with traditional edge detection methods, they are more sensitive to linear structures and enhance the continuity of weak targets.

[0030] For example, the gradient field magnitude in the horizontal z-direction Divide by the total horizontal gradient field amplitude thg Obtain the quotient, substitute it into the arctangent function to get the tilt angle. Substitute the tilt angle into the zero-crossing detection function to calculate the position where the tilt angle changes from positive to negative or from negative to positive, generating a binary image of the tilt angle zero-value line as the tilt angle zero-value line feature. This can be calculated as follows: Tilt =atan( / thg ); Where atan is the arctangent function. Tilt The angle of inclination.

[0031] Fea2=ZeroPN( Tilt )= ; The ZeroPN function is a zero-crossing detection function that determines... Tilt From the position where it changes from positive to negative or from negative to positive, Fea2 is characterized by a zero-value line of inclination angle. This refers to the zero-value line point in the zero-value line feature of the tilt angle.

[0032] The tilt angle zero-value line feature uses the boundary zero intersection point to delineate the edge contour. It does not depend on the gradient field amplitude change, but only on the ratio sign change. Compared with traditional methods, it has the ability to enhance the boundary of weak signal region and can effectively capture weak edges that are easy to miss.

[0033] For example, a window of size r×r is set, satisfying r < u and r < v. A pixel (i,j) is selected, and this pixel is used as the center point of the window. The sum of all pixels within the window is then divided by the window size to obtain the mean. The standard deviation is obtained by summing the squares of all pixels within the window after subtracting the mean, dividing by the window size, and then taking the square root of the result. Using the mean and standard deviation Set high and low thresholds. Preferably, set the high threshold to the mean plus 1.5 times the standard deviation, and the low threshold to the mean plus 0.5 times the standard deviation. If the total horizontal gradient field amplitude of the pixel at the center point is greater than the high threshold of the window, then the pixel at the center point is a strong edge point; if the total horizontal gradient field amplitude of the pixel at the center point is less than the high threshold of the window but greater than or equal to the low threshold of the window, then the pixel at the center point is a weak edge point, and the pixel is marked with a judgment result. Return to the step of selecting a pixel at the total horizontal gradient field amplitude until all pixels in the total horizontal gradient field amplitude are marked with a judgment result; generate an adaptive Canny edge feature map using the judgment results of all pixels; use the generated adaptive Canny edge feature map as the adaptive Canny feature. It can be calculated in the following way: ; in This represents the number of pixels contained in the window. For window size, This is the average value of all pixels within the window.

[0034] ; in This represents the number of pixels contained in the window. For window size, The average value of all pixels within the window. This represents the standard deviation of all pixels within the window.

[0035] T high_local = μ local +1.5 σ local ; T low_local = μ local +0.5 σ local ; in This is the average value of all pixels within the window. This represents the standard deviation of all pixels within the window. T high_local The high threshold of the window, T low_local This is the low threshold of the window.

[0036] Fea3= ; Fea3 is an adaptive Canny edge feature. These are edge points in the adaptive Canny edge features.

[0037] Traditional Canny edge detection relies on the selection of high and low thresholds, and a fixed threshold cannot adapt to changes in local image contrast. In areas of high contrast, a suitable threshold can filter out noise; in areas of low contrast, the same threshold may miss true edges, resulting in a high false alarm rate for edge recognition. This embodiment uses adaptive Canny edge features to detect local contrast. By dynamically adjusting the threshold to adapt to different edge strengths, it balances noise suppression with weak edge preservation, effectively reducing the false alarm rate of edge recognition.

[0038] Step 140: Fuse the multi-source contour features to obtain the initial edge contour.

[0039] This embodiment fuses the ridge feature, the zero-value tilt angle line feature, and the adaptive Canny feature to obtain an initial edge contour that simultaneously possesses the characteristics of all three contour features, achieving a synergistic effect. Optionally, the three features obtained above are all represented as matrices, and can be fused by addition to obtain the initial edge contour. M xy It can be calculated in the following way: M xy = ; in M xy For the initial edge contour, The ridge point in the ridge feature. The zero-value line point in the zero-value line feature of the tilt angle. These are edge points in the adaptive Canny edge features.

[0040] Step 150: Morphological processing is applied to the initial edge contour to obtain an enhanced edge contour.

[0041] In this embodiment, the initial edge contour Mxy obtained by the above steps may have outliers, breaks, and gaps.

[0042] Optionally, to remove outliers, edge deviance values ​​are applied to the initial edge contour. The size is set to r. A window of size r such that r < r And r < .right M xy For each pixel on an edge, take its neighborhood and calculate the mean of all pixels within that neighborhood. and standard deviation The absolute value of subtracting the mean from the pixels in the initial edge contour is compared with three times the standard deviation. If the absolute value of subtracting the mean from the pixels in the initial edge contour is greater than three times the standard deviation, the element is determined to be an outlier, and the outlier is replaced with the mean of its neighborhood. This can be calculated as follows: Z ij - >3• ; Z ij The element at coordinate (i, j) of the initial edge contour. and represents the mean and standard deviation of all pixels in the neighborhood.

[0043] Optionally, to address breakpoint and gap issues, breakpoint values ​​are connected to the initial edge contour. Based on the edge width, a radius is selected. d circle ， satisfy d＜ / 2 and d＜ / 2 To fill the small gap, an expansion operation is performed: ; in This is a morphological expansion function. The edge contour after expansion processing; The edge width is then restored through an erosion operation: ; in For morphological erosion functions, It is an enhanced edge contour obtained through morphological operations.

[0044] Step 160: After smoothing the enhanced edge contour, embed it into the original image to obtain the image with enhanced target edge.

[0045] Optionally, the obtained enhanced edge contours are smoothed using a Gaussian kernel and then embedded into the original image to output the enhanced target edge image. It can be calculated in the following way: ; in G (·) represents the Gaussian kernel. Preferably, the standard deviation of the Gaussian kernel can be set to 1.5. It is the target edge contour. It is an image after edge enhancement.

[0046] Compared to traditional image edge detection methods, the image edge enhancement method in this embodiment fuses ridge features, tilt angle zero-value features, and adaptive Canny edge features. Ridge features depict edge contours from the perspective of a linear geometric skeleton, resulting in lines with strong continuity; tilt angle zero-value features delineate edge contours using boundary zero-intersection points, achieving a high capture rate for weak signals; and adaptive Canny edge features generate edge contours using local contrast abrupt changes, resulting in a low false alarm rate for edge recognition. The initial edge contour obtained by fusing these three contour features possesses the advantages of all three, more effectively improving the edge response of weak targets and small anomalies, and reducing the false alarm rate.

[0047] Example 2 Figure 2 This is a schematic diagram of the image edge enhancement device provided in Embodiment 2 of the present invention. See also... Figure 2 The image edge enhancement device includes: The data acquisition module 210 is used to acquire the gravity gradient signal of the original image and obtain standardized gravity gradient tensor data by calculating and preprocessing the gravity gradient signal. The gradient calculation module 220 is used to calculate the normalized gravity gradient tensor data to obtain the gradient field amplitude, which includes: the gradient field amplitude in the horizontal direction, the gradient field amplitude in the vertical direction, the gradient field amplitude in the depth direction, and the total horizontal gradient field amplitude. Feature extraction module 230 is used to calculate and extract multi-source contour features from the gradient field amplitude. The calculation and extraction of multi-source contour features from the gradient field amplitude includes: extracting ridge features, which depict the edge contour from the perspective of a linear geometric skeleton; extracting tilt angle zero-value line features, which delineate the edge contour using boundary zero intersection points; and extracting adaptive Canny edge features, which generate the edge contour using local contrast abrupt changes. The fusion module 240 is used to fuse the multi-source contour features to obtain an initial edge contour; The morphological processing module 250 is used to perform morphological processing on the initial edge contour to obtain an enhanced edge contour. The output module 260 is used to smooth the enhanced edge contour and embed it into the original image to obtain an image with enhanced target edge.

[0048] The image edge enhancement device provided in this embodiment obtains standardized gravity gradient tensor data by acquiring and preprocessing the gravity gradient signal of the original image. Based on this data, it calculates the horizontal gradient field amplitude, vertical gradient field amplitude, depth gradient field amplitude, and total horizontal gradient field amplitude. Then, it extracts ridge features, tilt angle zero-value line features, and adaptive Canny edge features from the standardized gravity gradient tensor data and the gradient amplitudes, respectively. The three features are added pixel-level to obtain an initial edge contour, which is then smoothed using morphological processing to obtain an enhanced edge contour. Finally, the enhanced edge contour is embedded into the original image to output the edge-enhanced image. Ridge features depict the edge contour from the perspective of a linear geometric skeleton, resulting in lines with strong continuity. Tilt angle zero-value line features use boundary zero-intersection points to delineate the edge contour, achieving a high capture rate for weak signals. Adaptive Canny edge features generate edge contours using local contrast abrupt changes, resulting in a low false alarm rate for edge recognition. The initial edge contour obtained by fusing the three contour features possesses the advantages of all three, effectively improving the edge response of weak targets and small anomalies, and reducing the false alarm rate.

[0049] Based on the above embodiments, the data acquisition module includes: A two-dimensional gravity gradient data output unit is used to acquire the gravity position in the gravity gradient signal of the original image, and to obtain the two-dimensional gravity gradient data by performing a second-order partial derivative on the gravity position. The denoising unit is used to denoise the two-dimensional gravity gradient data and standardize the denoised two-dimensional gravity gradient data to obtain standardized gravity gradient tensor data.

[0050] Based on the above embodiments, the feature extraction module includes: The ridge point determination unit is used to determine whether a pixel is a ridge point by using the feature value of the standardized gravity gradient tensor data at each pixel point. The ridge intensity map generation unit is used to generate a ridge intensity map using the determination results of all pixels. A ridge feature output unit is used to use the ridge intensity map as the ridge feature.

[0051] Based on the above embodiments, the feature extraction module includes: A tilt angle output unit is used to obtain the tilt angle by using the ratio of the depth direction gradient field amplitude to the total horizontal gradient field amplitude; The tilt angle zero-value line plot generation unit is used to input the tilt angle into the zero-crossing detection function to generate a tilt angle zero-value line plot; The tilt angle zero-value line feature output unit is used to output the tilt angle zero-value line map as a tilt angle zero-value line feature.

[0052] Based on the above embodiments, the feature extraction module includes: A window setting unit is used to set a window of a fixed size, the window size of which should be smaller than the original image size; A selection unit is used to select a pixel on the total horizontal gradient field amplitude. A center point determination unit is used to determine the pixel as the center point of the window. The calculation unit is used to calculate the mean and standard deviation of all pixels within the window; A threshold output unit is used to derive the high threshold and low threshold of the window using the mean and the standard deviation; An edge point determination unit is used to determine whether the center point is a strong edge point or a weak edge point using the high threshold and low threshold of the window, and to mark the determination result on the pixel point that is the center point. The return unit is used to return the step of selecting a pixel on the total horizontal gradient field amplitude until all pixels on the total horizontal gradient field amplitude are marked with the judgment result; The adaptive Canny edge feature map generation unit is used to generate an adaptive Canny edge feature map using the judgment results of all pixels. An adaptive Canny edge feature output unit is used to use the adaptive Canny edge feature map as adaptive Canny edge features.

[0053] Based on the above embodiments, the fusion module includes: The fusion unit is used to directly add the ridge features, the tilt angle zero-value line features, and the Canny edge features to obtain the fused edge contour.

[0054] Based on the above embodiments, the morphological processing module includes: The processing unit is used to perform deviator value processing on the initial edge contour according to the 3σ criterion to obtain the edge contour after deviator value processing; A connection unit is used to connect the edge contours after the extinction value processing using dilation and erosion functions to obtain the reinforced edge contours.

[0055] The image edge enhancement device provided in this embodiment of the invention can execute the image edge enhancement method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0056] Example 3 Figure 3This is a schematic diagram of a device according to Embodiment 3 of the present invention. The device may be a general-purpose computer, an embedded image processing board, or a proprietary integrated circuit, and may include one or more processors and a memory. Figure 3 A block diagram of an exemplary device 12 suitable for implementing embodiments of the present invention is shown. Figure 3 The device 12 shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0057] like Figure 3 As shown, device 12 is represented as a general-purpose computing device. Components of device 12 may include, but are not limited to: one or more processors or processing units 16, system memory 28, and a bus 18 connecting different device components (including system memory 28 and processing unit 16).

[0058] Bus 18 represents one or more of several bus architectures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of the various bus architectures. For example, these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0059] Device 12 typically includes a variety of computer device readable media. These media can be any available media that can be accessed by device 12, including volatile and non-volatile media, removable and non-removable media.

[0060] System memory 28 may include computer device readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache 32. Device 12 may further include other removable / non-removable, volatile / non-volatile computer device storage media. By way of example only, storage device 34 may be used to read and write non-removable, non-volatile magnetic media (…). Figure 3 Not shown; usually referred to as a "hard drive"). Although Figure 3 Not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk") and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 18 via one or more data media interfaces. System memory 28 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the embodiments of the present invention.

[0061] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in system memory 28. Such program modules 42 include, but are not limited to, operating devices, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. Program modules 42 typically perform the functions and / or methods described in the embodiments of the present invention.

[0062] Device 12 can also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, etc.), and with one or more devices that enable a user to interact with device 12, and / or with any device that enables device 12 to communicate with one or more other computing devices (e.g., network card, modem, etc.). This communication can be performed via input / output (I / O) interface 22. Furthermore, device 12 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 20. Figure 3 As shown, network adapter 20 communicates with other modules of device 12 via bus 18. It should be understood that, although... Figure 3 As not shown, other hardware and / or software modules may be used in conjunction with device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID devices, tape drives, and data backup storage devices.

[0063] The processing unit 16 executes various functional applications and data processing by running programs stored in the system memory 28, such as implementing the image edge enhancement method provided in the embodiments of the present invention.

[0064] Example 4 Embodiment 4 of the present invention also provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform any of the image edge enhancement methods provided in the above embodiments.

[0065] The computer storage medium of this invention can be any combination of one or more computer-readable media. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. For example, a computer-readable storage medium can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used or combined with an instruction execution device, apparatus, or device.

[0066] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit programs for use by or in connection with an instruction execution device, apparatus, or apparatus.

[0067] Program code contained on a computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0068] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or device. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0069] Note that the above description is merely a preferred embodiment of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of the present invention, the scope of which is determined by the scope of the appended claims.

Claims

1. An image edge enhancement method characterized by, include: The gravity gradient signal of the original image is obtained, and the normalized gravity gradient tensor data is obtained by calculating and preprocessing the gravity gradient signal. The normalized gravity gradient tensor data is used to calculate the gradient field amplitude, which includes: the gradient field amplitude in the horizontal direction, the gradient field amplitude in the vertical direction, the gradient field amplitude in the depth direction, and the total horizontal gradient field amplitude. Calculating and extracting multi-source contour features from the gradient field amplitude, wherein the calculation and extraction of multi-source contour features from the gradient field amplitude includes: Extract ridge features, which depict edge contours from the perspective of a linear geometric skeleton; Extract the zero-value line feature of the tilt angle, and use the boundary zero intersection point to delineate the edge contour; Extract adaptive Canny edge features, which generate edge contours using local contrast abrupt changes; The initial edge contour is obtained by fusing the multi-source contour features; The initial edge contour is morphologically processed to obtain an enhanced edge contour; The enhanced edge contour is smoothed and then embedded into the original image to obtain an image with enhanced target edges.

2. The method of claim 1, wherein, The process of acquiring the gravity gradient signal from the original image and obtaining standardized gravity gradient tensor data through calculation and preprocessing of the gravity gradient signal includes: The gravity position in the gravity gradient signal of the original image is obtained, and the second-order partial derivative of the gravity position is performed to obtain the two-dimensional gravity gradient data. The two-dimensional gravity gradient data is denoised, and the denoised two-dimensional gravity gradient data is standardized to obtain standardized gravity gradient tensor data.

3. The method according to claim 1, characterized in that, The extraction of ridge features includes: The feature values ​​of the standardized gravity gradient tensor data at each pixel are used to determine whether the pixel is a ridge point; Generate a ridge intensity map using the determination results of all pixels; The ridge intensity map is used as the ridge feature.

4. The method according to claim 1, characterized in that, The extraction of the zero-value line feature of the tilt angle includes: The tilt angle is obtained by using the ratio of the depth-direction gradient field amplitude to the total horizontal gradient field amplitude; Substitute the tilt angle into the zero-cross detection function to generate a tilt angle zero-value line graph; The zero-value line diagram of the tilt angle is used as the feature of the zero-value line of the tilt angle.

5. The method according to claim 1, characterized in that, The extraction of adaptive Canny edge features includes: Set a window of a fixed size, which should be smaller than the size of the original image; Select a pixel on the total horizontal gradient field amplitude; The pixel is used as the center point of the window; Calculate the mean and standard deviation of all pixels within the window; The high and low thresholds of the window are derived using the mean and the standard deviation. The high and low thresholds of the window are used to determine whether the center point is a strong edge point or a weak edge point, and the determination result is marked on the pixel point that is the center point. Return to the step of selecting a pixel on the total horizontal gradient field amplitude, until all pixels on the total horizontal gradient field amplitude are marked with the judgment result; An adaptive Canny edge feature map is generated using the judgment results of all pixels; The adaptive Canny edge feature map is used as the adaptive Canny edge feature.

6. The method according to claim 1, characterized in that, The process of fusing the multi-source contour features to obtain the fused edge contour includes: The ridge features, the zero-value tilt angle features, and the Canny edge features are directly added together to obtain the fused edge contour.

7. The method according to claim 1, characterized in that, The morphological processing of the initial edge contour to obtain an enhanced edge contour includes: The initial edge contour is subjected to derivation value processing according to the 3σ criterion to obtain the edge contour after derivation value processing; The edge contour after the extinction value processing is connected by breaking points using dilation and erosion functions to obtain the enhanced edge contour.

8. An image edge enhancement device, characterized in that, include: The data acquisition module is used to acquire the gravity gradient signal of the original image, and to obtain standardized gravity gradient tensor data by calculating and preprocessing the gravity gradient signal. The gradient calculation module is used to calculate the horizontal gradient field amplitude, the vertical gradient field amplitude, the depth gradient field amplitude, and the total horizontal gradient field amplitude based on the standardized gravity gradient tensor data. The feature extraction module is used to extract multi-source contour features from the gradient field amplitude: ridge features, which depict the edge contour from the perspective of a linear geometric skeleton; tilt angle zero-value line features, which delineate the edge contour using boundary zero intersection points; and adaptive Canny edge features, which generate the edge contour using local contrast abrupt changes. The fusion module is used to fuse the multi-source contour features to obtain an initial edge contour; The morphological processing module is used to perform morphological processing on the initial edge contour to obtain an enhanced edge contour. The output module is used to embed the enhanced edge contour into the original image to obtain an image with enhanced target edges.

9. A device, characterized in that, include: One or more processors; Storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the image edge enhancement method as described in claims 1-7.

10. A storage medium containing computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a computer processor, are used to perform the image edge enhancement method as described in any one of claims 1-7.

Images