Positioning precision improving method and system based on machine vision

Abstract

The invention relates to the technical field of industrial visual positioning, and provides a positioning precision improvement method and system based on machine vision, and the method comprises the steps: carrying out the fusion processing of multi-view image information based on an image fusion algorithm, and obtaining a fused image; performing sub-pixel edge detection processing on the fused image to obtain a target image; determining the pose of the to-be-positioned target based on the target image; obtaining a standard pose of the to-be-positioned target, and generating a positioning scheme of the to-be-positioned target based on the standard pose and the pose; and positioning the to-be-positioned target based on the positioning scheme, acquiring motion parameter information of production line equipment where the to-be-positioned target is located in real time in the process of positioning the to-be-positioned target, and adjusting the position of the to-be-positioned target in real time based on the motion parameter information. According to the method, the workpiece can be accurately positioned without depending on a fixed calibration device or a laser range finder.

Description

Technical Field

[0001] The present application relates to the field of industrial visual positioning technology, and in particular to a positioning accuracy improvement method and system based on machine vision. Background Art

[0002] In the process of automated industrial production, precise positioning technology is the key to achieving efficient operation of automated production lines. Traditional positioning technologies mostly rely on fixed calibration devices or laser rangefinders. These positioning solutions have shown the characteristics of high accuracy and stability. However, they usually face challenges such as high cost, complex installation and maintenance, and susceptibility to interference from the external environment. Summary of the invention

[0003] The present application provides a method and system for improving positioning accuracy based on machine vision to solve the problems raised by the above background technology.

[0004] In a first aspect, the present application provides a method for improving positioning accuracy based on machine vision, comprising: Acquire multi-view image information of the target to be located, and fuse the multi-view image information based on a preset image fusion algorithm to obtain a fused image; Performing sub-pixel edge detection processing on the fused image to obtain a target image; Determining the position and posture of the target to be located based on the target image; Acquire a standard posture of the target to be located in a database, and generate a positioning solution for the target to be located based on the standard posture and the posture; The target to be located is positioned based on the positioning scheme, and during the positioning process of the target to be located, motion parameter information of the production line equipment where the target to be located is obtained in real time, and the position of the target to be located is adjusted in real time based on the motion parameter information.

[0005] In a possible implementation, the fusing the multi-view image information based on a preset image fusion algorithm to obtain a fused image includes: Acquire the clarity of each image in the multi-view image information based on a preset image clarity acquisition method, and generate a weight of each image based on the clarity of each image; Performing spatial domain fusion processing on the multi-view image information based on the weights of the images and a preset spatial domain fusion algorithm to obtain a spatial domain image; Performing frequency domain fusion processing on the multi-view image information based on a preset frequency domain fusion algorithm to obtain a frequency domain image; The spatial domain image and the frequency domain image are fused to obtain the fused image.

[0006] In a possible implementation manner, generating the weight of each of the images based on the definition of each of the images includes: Adding the sharpness of each of the images to obtain a sum of sharpness; For each of the images, the percentage of the definition of the image to the sum of the definitions is determined as the weight of the image.

[0007] In a possible implementation, performing spatial domain fusion processing on the multi-view image information based on the weights of the images and a preset spatial domain fusion algorithm to obtain a spatial domain image includes: For each image in the multi-view image information, perform Laplace pyramid decomposition on the image based on a preset Laplace pyramid decomposition algorithm to obtain a Laplace pyramid of the image; the Laplace pyramid includes a plurality of Gaussian layers and a plurality of Laplace layers; The Laplacian pyramids of the images are fused based on the weights of the images to obtain a fused Laplacian pyramid; wherein the fused Laplacian pyramid includes a plurality of fused Gaussian layers and a plurality of fused Laplacian layers; Each layer of the fused Laplacian pyramid is fused to obtain the spatial domain image.

[0008] In a possible implementation manner, performing Laplace pyramid decomposition on the image based on a preset Laplace pyramid decomposition algorithm to obtain a Laplace pyramid of the image includes: Performing Gaussian blur processing and downsampling processing on the image for a preset number of times successively to generate a set of Gaussian blurred images arranged in order from bottom to top with gradually decreasing resolution and size; The Laplacian layer is inserted between each two adjacent Gaussian blurred images to obtain the Laplacian pyramid.

[0009] In a possible implementation, performing frequency domain fusion processing on the multi-view image information based on a preset frequency domain fusion algorithm to obtain a frequency domain image includes: Performing two-dimensional wavelet decomposition on each image in the multi-view image information to obtain wavelet coefficient information corresponding to each image; wherein the wavelet coefficient information includes low-frequency coefficients, horizontal high-frequency coefficients, vertical high-frequency coefficients and diagonal high-frequency coefficients; Calculate the average value of each wavelet coefficient information to obtain the low-frequency fusion coefficient; Determine the maximum horizontal high frequency coefficient among the horizontal high frequency coefficients as the target horizontal high frequency coefficient; Determine the maximum vertical high-frequency coefficient among the vertical high-frequency coefficients as the target vertical high-frequency coefficient; Determine the maximum diagonal high-frequency coefficient among the diagonal high-frequency coefficients as the target diagonal high-frequency coefficient; The low-frequency fusion coefficient, the target horizontal high-frequency coefficient, the target vertical high-frequency coefficient and the target diagonal high-frequency coefficient are subjected to inverse wavelet transformation to obtain the frequency domain image.

[0010] In a possible implementation, performing sub-pixel edge detection processing on the fused image to obtain a target image includes: Based on a preset image gradient algorithm, respectively obtain the horizontal gradient and the vertical gradient of each pixel point of the target image; For each pixel of the target image, obtaining a gradient magnitude of the pixel based on a horizontal gradient and a vertical gradient of the pixel, comparing the gradient magnitude with a preset gradient threshold, and determining that the pixel is an edge point when the gradient magnitude is greater than the gradient threshold; For each edge point, determining the gradient direction of the edge point based on the horizontal gradient and the vertical gradient of the edge point, and correcting the gradient direction based on a preset gradient direction correction method to obtain the main gradient direction of the edge point; For each edge point, two adjacent pixel points of the edge point are determined in the main gradient direction of the edge point, and the gradient amplitude of the edge point is compared with the gradient amplitude of each adjacent pixel point respectively, and when the gradient amplitude of the edge point is greater than the gradient amplitude of each adjacent pixel point, the edge point is determined to be a target edge point; For each of the target edge points, using a quadratic difference method to determine a sub-pixel point of the target edge point; Based on each of the sub-pixel points, each of the target edge points is connected using a preset edge tracking algorithm to obtain the target image.

[0011] In a second aspect, the present application provides a positioning accuracy improvement system based on machine vision, comprising: An acquisition module is used to acquire multi-view image information of the target to be located, and fuse the multi-view image information based on a preset image fusion algorithm to obtain a fused image; An edge detection processing module is used to perform sub-pixel edge detection processing on the fused image to obtain a target image; A determination module, used to determine the position and posture of the target to be located based on the target image; A generation module, which obtains the standard posture of the target to be located in a database, and generates a positioning solution for the target to be located based on the standard posture and the posture; The adjustment module is used to locate the target to be located based on the positioning scheme, and in the process of locating the target to be located, obtain the motion parameter information of the production line equipment where the target to be located is located in real time, and adjust the position of the target to be located in real time based on the motion parameter information.

[0012] The present application provides a method and system for improving positioning accuracy based on machine vision, the method comprising: obtaining multi-view image information of a target to be located, and fusing the multi-view image information based on a preset image fusion algorithm to obtain a fused image; performing sub-pixel edge detection on the fused image to obtain a target image; determining the position and posture of the target to be located based on the target image; obtaining a standard position and posture of the target to be located in a database, and generating a positioning scheme for the target to be located based on the standard position and the position; performing positioning processing on the target to be located based on the positioning scheme, and obtaining motion parameter information of a production line equipment where the target to be located is located in real time during the positioning processing of the target to be located, and adjusting the position of the target to be located in real time based on the motion parameter information. The method, on the one hand, improves the positioning accuracy through machine vision technology, realizes accurate positioning of the workpiece without relying on fixed calibration devices or laser rangefinders, reduces economic costs, and the complexity of installation and maintenance; on the other hand, improves the positioning accuracy through image fusion technology; on still another hand, by acquiring the motion parameter information of the production line equipment where the target to be located is located in real time during the positioning process of the target to be located, and adjusting the position of the target to be located in real time based on the motion parameter information, adaptive adjustment of the positioning scheme is realized, and the positioning accuracy is further improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.

[0014] Figure 1 A flowchart of a method for improving positioning accuracy based on machine vision provided in an embodiment of the present application; Figure 2 A schematic block diagram of the structure of a positioning accuracy improvement system based on machine vision provided in an embodiment of the present application; Figure 3 A schematic block diagram of the structure of a terminal device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0015] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0016] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0017] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0018] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0019] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0020] See also Figure 1 , Figure 1 A flowchart of a method for improving positioning accuracy based on machine vision provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the method for improving positioning accuracy based on machine vision provided in the embodiment of the present application includes steps S1 to S5.

[0021] Step S1: Acquire multi-view image information of a target to be located, and fuse the multi-view image information based on a preset image fusion algorithm to obtain a fused image.

[0022] Specifically, the image of the target to be located is collected by each camera device arranged at a preset position to obtain the multi-view image information, and the multi-view image information is fused based on a preset image fusion algorithm to obtain a fused image, including the following steps: Based on a preset image clarity acquisition method, the clarity of each image in the multi-view image information is respectively acquired, and the weight of each image is generated based on the clarity of each image; wherein the image clarity acquisition method is any one of a gradient amplitude method, a Laplace operator energy method, an edge density method, and a high-frequency component energy method; Performing spatial domain fusion processing on the multi-view image information based on the weights of the images and a preset spatial domain fusion algorithm to obtain a spatial domain image; Performing frequency domain fusion processing on the multi-view image information based on a preset frequency domain fusion algorithm to obtain a frequency domain image; The spatial domain image and the frequency domain image are fused to obtain the fused image; specifically, weighted sum of corresponding pixels of the spatial domain image and the frequency domain image is performed to obtain the fused image, and the weights of the spatial domain image and the frequency domain image are both 0.5.

[0023] It can be understood that the above-mentioned method of fusing the multi-view image information based on a preset image fusion algorithm to obtain a fused image improves the detail display effect of the image by fusing multi-view and multi-domain images, thereby providing a reliable data basis for subsequent visual detection and positioning processing.

[0024] The step of generating the weight of each image based on the definition of each image comprises the following steps: Adding the sharpness of each of the images to obtain a sum of sharpness; For each of the images, the percentage of the definition of the image to the sum of the definitions is determined as the weight of the image.

[0025] The performing spatial domain fusion processing on the multi-view image information based on the weights of each of the images and a preset spatial domain fusion algorithm to obtain a spatial domain image comprises the following steps: For each image in the multi-view image information, a Laplace pyramid decomposition is performed on the image based on a preset Laplace pyramid decomposition algorithm to obtain a Laplace pyramid of the image; the Laplace pyramid includes multiple Gaussian layers and multiple Laplace layers; specifically, the Laplace pyramid decomposition is performed on the image based on the preset Laplace pyramid decomposition algorithm to obtain the Laplace pyramid of the image, including: performing Gaussian blur processing and downsampling processing on the image for a preset number of times, generating a group of Gaussian blurred images with gradually decreasing resolution and size arranged in order from bottom to top; inserting Laplace layers between each two adjacent Gaussian blurred images to obtain the Laplace pyramid; wherein the bottom layer of the Laplace pyramid is the image, and the Gaussian blur processing and downsampling processing on the image for a preset number of times includes: performing Gaussian blur processing on the image, and performing interval deletion on each pixel of the image after the Gaussian blur processing, and splicing the remaining pixels to obtain a first Gaussian Blurred image layer; performing Gaussian blur processing on the first Gaussian blurred image layer, and performing interval deletion on each pixel of the first Gaussian blurred image layer after the Gaussian blur processing, and splicing the remaining pixels to obtain a second Gaussian blurred image layer; successively superimposing until a preset number of Gaussian blur processing and downsampling processing are completed; the Laplace layer is inserted between each adjacent two Gaussian blurred images, including: for two adjacent Gaussian blurred images, using a preset bilinear difference method to upsample the upper Gaussian blurred image so that the upper Gaussian blurred image and the lower Gaussian blurred image have the same number of pixels, and performing a difference calculation of corresponding pixels between the lower Gaussian blurred image and the upper Gaussian blurred image after the upsampling processing to obtain the Laplace layer between the two adjacent Gaussian blurred images; it should be noted that the method of performing a difference calculation of corresponding pixels between the lower Gaussian blurred image and the upper Gaussian blurred image after the upsampling processing is to use the pixels of the lower Gaussian blurred image to subtract the corresponding pixels of the upper Gaussian blurred image after the upsampling processing; The Laplacian pyramids of the images are fused based on the weights of the images to obtain a fused Laplacian pyramid; wherein the fused Laplacian pyramid includes a plurality of fused Gaussian layers and a plurality of fused Laplacian layers; specifically, based on the weights of the images corresponding to the Laplacian pyramids, corresponding pixels of the same layer of the Laplacian pyramids are weighted summed to obtain the fused Laplacian pyramid; The layers of the fused Laplacian pyramid are fused to obtain the spatial domain image. Specifically, the fusion processing of the layers of the fused Laplacian pyramid to obtain the spatial domain image includes: for each fused Laplacian layer, superimposing the fused Laplacian layer with the fused Gaussian layer above it by corresponding pixels to obtain an initial target fusion layer; upsampling the initial target fusion layer of the highest layer based on a preset bilinear difference method so that the initial target fusion layer of the highest layer has the same number of pixels as the initial target fusion layer below it, and superimposing the initial target fusion layer of the highest layer after the upsampling processing with the initial target fusion layer below it by corresponding pixels to obtain a first intermediate target fusion layer; upsampling the first intermediate target fusion layer based on a preset bilinear difference method so that the first intermediate target fusion layer has the same number of pixels as the initial target fusion layer below it, and superimposing the first intermediate target fusion layer after the upsampling processing with the initial target fusion layer below it by corresponding pixels to obtain a second intermediate target fusion layer; and superimposing successively until each of the initial target fusion layers is fused to the bottom layer of the fused Laplacian pyramid.

[0026] It can be understood that the method of performing spatial domain fusion processing on the multi-view image information based on the weights of each image and a preset spatial domain fusion algorithm to obtain a spatial domain image, by obtaining the Laplacian pyramid of each image, and fusing the Laplacian pyramids of each image based on the weights of each image to obtain a fused Laplacian pyramid, and fusing each layer of the fused Laplacian pyramid to obtain the spatial domain image, can accurately extract the high-frequency detail information of the image, enhance the clarity and visual details of the image, improve the quality of the fused image, and thereby improve the positioning accuracy.

[0027] The method of performing frequency domain fusion processing on the multi-view image information based on a preset frequency domain fusion algorithm to obtain a frequency domain image includes the following steps: Performing two-dimensional wavelet decomposition on each image in the multi-view image information to obtain wavelet coefficient information corresponding to each image; wherein the wavelet coefficient information includes low-frequency coefficients, horizontal high-frequency coefficients, vertical high-frequency coefficients and diagonal high-frequency coefficients; Calculate the average value of each wavelet coefficient information to obtain the low-frequency fusion coefficient; Determine the maximum horizontal high frequency coefficient among the horizontal high frequency coefficients as the target horizontal high frequency coefficient; Determine the maximum vertical high-frequency coefficient among the vertical high-frequency coefficients as the target vertical high-frequency coefficient; Determine the maximum diagonal high-frequency coefficient among the diagonal high-frequency coefficients as the target diagonal high-frequency coefficient; The low-frequency fusion coefficient, the target horizontal high-frequency coefficient, the target vertical high-frequency coefficient and the target diagonal high-frequency coefficient are subjected to inverse wavelet transformation to obtain the frequency domain image.

[0028] It can be understood that the above-mentioned method of performing frequency domain fusion processing on the multi-view image information based on a preset frequency domain fusion algorithm to obtain a frequency domain image highly integrates the useful information in the multi-view image through wavelet transform, which helps to further improve the quality and information expression of the fused image, thereby improving the positioning accuracy.

[0029] Step S2: performing sub-pixel edge detection processing on the fused image to obtain a target image.

[0030] Specifically, step S2 includes the following steps: Based on a preset image gradient algorithm, the horizontal gradient and the vertical gradient of each pixel point of the target image are respectively obtained; wherein the image gradient algorithm is a Sobel operator or a Prewitt operator; For each pixel of the target image, a gradient amplitude of the pixel is obtained based on the horizontal gradient and the vertical gradient of the pixel, and the gradient amplitude is compared with a preset gradient threshold, and when the gradient amplitude is greater than the gradient threshold, the pixel is determined to be an edge point; specifically, the method of obtaining the gradient amplitude of the pixel based on the horizontal gradient and the vertical gradient of the pixel is to calculate the square sum of the horizontal gradient and the vertical gradient, and determine the positive square root of the square sum as the gradient amplitude; For each edge point, the gradient direction of the edge point is determined based on the horizontal gradient and the vertical gradient of the edge point, and the gradient direction is corrected based on a preset gradient direction correction method to obtain the main gradient direction of the edge point; specifically, the gradient direction of the edge point is determined using formula (1). It should be noted that four main gradient directions of 45°, 90°, 135°, and 0° are usually included in edge detection. Correcting the gradient direction based on a preset gradient direction correction method to obtain the main gradient direction of the edge point means selecting the main gradient direction closest to the gradient direction in each main gradient direction as the main gradient direction of the edge point. For example, if the gradient direction is 38°, then the main gradient direction of the edge point is 45°.

[0031] in, θ(x,y) For location (x,y) The gradient direction of the edge point at Gy For location (x,y) The vertical gradient of the edge point at Gx For location (x,y)The horizontal gradient of the edge point at .

[0032] For each edge point, two adjacent pixel points of the edge point are determined in the main gradient direction of the edge point, and the gradient amplitude of the edge point is compared with the gradient amplitude of each adjacent pixel point, and when the gradient amplitude of the edge point is greater than the gradient amplitude of each adjacent pixel point, the edge point is determined to be a target edge point; illustratively, if the position of the edge point is (x, y) and the main gradient direction is 45°, the positions of the two adjacent pixel points are (x-1, y-1) and (x+1, y+1), respectively. If the edge point is located at (x, y) and the main gradient direction is 90°, the positions of the two adjacent pixels are (x, y-1) and (x, y+1) respectively. If the edge point is located at (x, y) and the main gradient direction is 135°, the positions of the two adjacent pixels are (x-1, y+1) and (x+1, y-1) respectively. If the edge point is located at (x, y) and the main gradient direction is 0°, the positions of the two adjacent pixels are (x-1, y) and (x+1, y) respectively. For each of the target edge points, using a quadratic difference method to determine a sub-pixel point of the target edge point; Based on each of the sub-pixel points, each of the target edge points is connected using a preset edge tracking algorithm to obtain the target image.

[0033] It can be understood that the above method of performing sub-pixel edge detection processing on the fused image to obtain the target image, from edge detection to sub-pixel positioning, and then to edge tracking, each step is aimed at improving the accuracy and reliability of edge detection effect, thereby helping to improve positioning accuracy.

[0034] Step S3: determining the position and posture of the target to be located based on the target image.

[0035] Specifically, the target image is input into a preset pose extraction model to obtain the pose of the target to be located, wherein the pose extraction model is a pre-trained deep learning model. It can be understood that the pose includes the position and posture of the target to be located.

[0036] Step S4: acquiring a standard posture of the target to be located in a database, and generating a positioning solution for the target to be located based on the standard posture and the posture.

[0037] Among them, the standard posture is the posture of the target to be positioned after the target to be positioned is accurately positioned on the production line equipment, and the method for obtaining the standard posture is to accurately position the workpiece consistent with the target to be positioned when the production line equipment is stationary, and then acquire the posture of the workpiece to obtain the standard posture.

[0038] Step S5: positioning the target to be positioned based on the positioning scheme, and obtaining in real time the motion parameter information of the production line equipment where the target to be positioned is located during the positioning process, and adjusting the position and posture of the target to be positioned in real time based on the motion parameter information.

[0039] Specifically, the motion parameter information is input into a preset posture deviation extraction model to obtain the posture deviation of the production line equipment relative to when it is stationary, and the posture of the target to be positioned is adjusted in real time based on the posture deviation. The posture deviation extraction model is a pre-trained neural network model.

[0040] The method provided in this embodiment, on the one hand, improves the positioning accuracy through machine vision technology, and realizes accurate positioning of the workpiece without relying on fixed calibration devices or laser rangefinders, thereby reducing economic costs and the complexity of installation and maintenance. On the other hand, it improves the positioning accuracy through image fusion technology. On the other hand, by acquiring the motion parameter information of the production line equipment where the target to be located is located in real time during the positioning process of the target to be located, and adjusting the position of the target to be located in real time based on the motion parameter information, it realizes adaptive adjustment of the positioning scheme and further improves the positioning accuracy.

[0041] See also Figure 2 , Figure 2 The structure schematic block diagram of the positioning accuracy improvement system 100 based on machine vision provided in the embodiment of the present application is as follows: Figure 2 As shown, the positioning accuracy improvement system 100 based on machine vision provided in the embodiment of the present application includes: The acquisition module 110 is used to acquire multi-view image information of the target to be located, and fuse the multi-view image information based on a preset image fusion algorithm to obtain a fused image.

[0042] The edge detection processing module 120 is used to perform sub-pixel edge detection processing on the fused image to obtain a target image.

[0043] The determination module 130 is used to determine the position and posture of the target to be located based on the target image.

[0044] The generating module 140 obtains the standard posture of the target to be located in the database, and generates a positioning solution for the target to be located based on the standard posture and the posture.

[0045] The adjustment module 150 is used to locate the target to be located based on the positioning scheme, and in the process of locating the target to be located, obtain the motion parameter information of the production line equipment where the target to be located is located in real time, and adjust the position of the target to be located in real time based on the motion parameter information.

[0046] It should be noted that technical personnel in the relevant technical field can clearly understand that, for the convenience and conciseness of description, the specific working process of the system and each module described above can refer to the corresponding process in the aforementioned embodiment of the positioning accuracy improvement method based on machine vision, and will not be repeated here.

[0047] The positioning accuracy improvement system 100 based on machine vision provided in the above embodiment can be implemented in the form of a computer program. The computer program can be used in Figure 3 The system is run on the terminal device 200 shown.

[0048] See also Figure 3 , Figure 3 The present invention provides a schematic block diagram of the structure of a terminal device 200 according to an embodiment of the present application. The terminal device 200 includes a processor 201 and a memory 202. The processor 201 and the memory 202 are connected via a device bus 203, wherein the memory 202 may include a non-volatile storage medium and an internal memory.

[0049] The non-volatile storage medium can store a computer program. The computer program includes program instructions, and when the program instructions are executed by the processor 201, the processor 201 can execute any of the above-mentioned positioning accuracy improvement methods based on machine vision.

[0050] The processor 201 is used to provide computing and control capabilities to support the operation of the entire terminal device 200.

[0051] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor 201, the processor 201 can execute any of the above-mentioned positioning accuracy improvement methods based on machine vision.

[0052] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a partial structure related to the scheme of the present application, and does not constitute a limitation on the terminal device 200 involved in the scheme of the present application. The specific terminal device 200 may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0053] It should be understood that the processor 201 may be a central processing unit (CPU), and the processor 201 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0054] In some embodiments, the processor 201 is used to run a computer program stored in the memory to implement the following steps: Acquire multi-view image information of the target to be located, and fuse the multi-view image information based on a preset image fusion algorithm to obtain a fused image; Performing sub-pixel edge detection processing on the fused image to obtain a target image; Determining the position and posture of the target to be located based on the target image; Acquire a standard posture of the target to be located in a database, and generate a positioning solution for the target to be located based on the standard posture and the posture; The target to be located is positioned based on the positioning scheme, and during the positioning process of the target to be located, motion parameter information of the production line equipment where the target to be located is obtained in real time, and the position of the target to be located is adjusted in real time based on the motion parameter information.

[0055] It should be noted that technical personnel in the relevant field can clearly understand that for the convenience and conciseness of description, the specific working process of the terminal device 200 described above can refer to the corresponding process of the aforementioned positioning accuracy improvement method based on machine vision, and will not be repeated here.

[0056] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by one or more processors, the one or more processors implement the positioning accuracy improvement method based on machine vision as provided in the embodiment of the present application.

[0057] The computer-readable storage medium may be an internal storage unit of the terminal device 200 in the aforementioned embodiment, such as a hard disk or memory of the terminal device 200. The computer-readable storage medium may also be an external storage device of the terminal device 200, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped with the terminal device 200.

[0058] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for improving positioning accuracy based on machine vision, characterized in that: include: Acquire multi-view image information of the target to be located, and fuse the multi-view image information based on a preset image fusion algorithm to obtain a fused image; Performing sub-pixel edge detection processing on the fused image to obtain a target image; Determining the position and posture of the target to be located based on the target image; Acquire a standard posture of the target to be located in a database, and generate a positioning solution for the target to be located based on the standard posture and the posture; The target to be located is positioned based on the positioning scheme, and during the positioning process of the target to be located, motion parameter information of the production line equipment where the target to be located is obtained in real time, and the position of the target to be located is adjusted in real time based on the motion parameter information.

2. The method for improving positioning accuracy based on machine vision according to claim 1, characterized in that: The fusing the multi-view image information based on a preset image fusion algorithm to obtain a fused image includes: Acquire the clarity of each image in the multi-view image information based on a preset image clarity acquisition method, and generate a weight of each image based on the clarity of each image; Performing spatial domain fusion processing on the multi-view image information based on the weights of the images and a preset spatial domain fusion algorithm to obtain a spatial domain image; Performing frequency domain fusion processing on the multi-view image information based on a preset frequency domain fusion algorithm to obtain a frequency domain image; The spatial domain image and the frequency domain image are fused to obtain the fused image.

3. The method for improving positioning accuracy based on machine vision according to claim 2, characterized in that: Generating the weight of each of the images based on the clarity of each of the images comprises: Adding the sharpness of each of the images to obtain a sum of sharpness; For each of the images, the percentage of the definition of the image to the sum of the definitions is determined as the weight of the image.

4. The method for improving positioning accuracy based on machine vision according to claim 2, characterized in that: The performing spatial domain fusion processing on the multi-view image information based on the weights of each of the images and a preset spatial domain fusion algorithm to obtain a spatial domain image includes: For each image in the multi-view image information, perform Laplace pyramid decomposition on the image based on a preset Laplace pyramid decomposition algorithm to obtain a Laplace pyramid of the image; the Laplace pyramid includes a plurality of Gaussian layers and a plurality of Laplace layers; The Laplacian pyramids of the images are fused based on the weights of the images to obtain a fused Laplacian pyramid; wherein the fused Laplacian pyramid includes a plurality of fused Gaussian layers and a plurality of fused Laplacian layers; Each layer of the fused Laplacian pyramid is fused to obtain the spatial domain image.

5. The method for improving positioning accuracy based on machine vision according to claim 4, characterized in that: The step of performing Laplace pyramid decomposition on the image based on a preset Laplace pyramid decomposition algorithm to obtain a Laplace pyramid of the image includes: Performing Gaussian blur processing and downsampling processing on the image for a preset number of times successively to generate a set of Gaussian blurred images arranged in order from bottom to top with gradually decreasing resolution and size; The Laplacian layer is inserted between each two adjacent Gaussian blurred images to obtain the Laplacian pyramid.

6. The method for improving positioning accuracy based on machine vision according to claim 2, characterized in that: The performing frequency domain fusion processing on the multi-view image information based on a preset frequency domain fusion algorithm to obtain a frequency domain image includes: Performing two-dimensional wavelet decomposition on each image in the multi-view image information to obtain wavelet coefficient information corresponding to each image; wherein the wavelet coefficient information includes low-frequency coefficients, horizontal high-frequency coefficients, vertical high-frequency coefficients and diagonal high-frequency coefficients; Calculate the average value of each wavelet coefficient information to obtain the low-frequency fusion coefficient; Determine the maximum horizontal high frequency coefficient among the horizontal high frequency coefficients as the target horizontal high frequency coefficient; Determine the maximum vertical high-frequency coefficient among the vertical high-frequency coefficients as the target vertical high-frequency coefficient; Determine the maximum diagonal high-frequency coefficient among the diagonal high-frequency coefficients as the target diagonal high-frequency coefficient; The low-frequency fusion coefficient, the target horizontal high-frequency coefficient, the target vertical high-frequency coefficient and the target diagonal high-frequency coefficient are subjected to inverse wavelet transformation to obtain the frequency domain image.

7. The method for improving positioning accuracy based on machine vision according to claim 1, characterized in that: The performing sub-pixel edge detection processing on the fused image to obtain a target image includes: Based on a preset image gradient algorithm, respectively obtain the horizontal gradient and the vertical gradient of each pixel point of the target image; For each pixel of the target image, obtaining a gradient magnitude of the pixel based on a horizontal gradient and a vertical gradient of the pixel, comparing the gradient magnitude with a preset gradient threshold, and determining that the pixel is an edge point when the gradient magnitude is greater than the gradient threshold; For each edge point, determining the gradient direction of the edge point based on the horizontal gradient and the vertical gradient of the edge point, and correcting the gradient direction based on a preset gradient direction correction method to obtain the main gradient direction of the edge point; For each edge point, two adjacent pixel points of the edge point are determined in the main gradient direction of the edge point, and the gradient amplitude of the edge point is compared with the gradient amplitude of each adjacent pixel point respectively, and when the gradient amplitude of the edge point is greater than the gradient amplitude of each adjacent pixel point, the edge point is determined to be a target edge point; For each of the target edge points, using a quadratic difference method to determine a sub-pixel point of the target edge point; Based on each of the sub-pixel points, each of the target edge points is connected using a preset edge tracking algorithm to obtain the target image.

8. A positioning accuracy improvement system based on machine vision, characterized in that: include: An acquisition module is used to acquire multi-view image information of the target to be located, and fuse the multi-view image information based on a preset image fusion algorithm to obtain a fused image; An edge detection processing module is used to perform sub-pixel edge detection processing on the fused image to obtain a target image; A determination module, used to determine the position and posture of the target to be located based on the target image; A generation module, which obtains the standard posture of the target to be located in a database, and generates a positioning solution for the target to be located based on the standard posture and the posture; The adjustment module is used to locate the target to be located based on the positioning scheme, and in the process of locating the target to be located, obtain the motion parameter information of the production line equipment where the target to be located is located in real time, and adjust the position of the target to be located in real time based on the motion parameter information.

Images