An intelligent measurement and analysis system for imaging instruments

Abstract

This invention relates to the field of intelligent measurement technology for imaging instruments, and specifically to an intelligent measurement and analysis system for imaging instruments. The invention acquires grayscale images of components on an industrial production line in real time using an imaging instrument. Motion blur dynamic compensation is applied to the grayscale images to generate initial component images. These initial images are then preprocessed to obtain target component images. Based on multi-feature fusion, grayscale gradient data, texture feature data, and spatial distribution feature data of the target component images are extracted. By comparing the grayscale images with a sample set of components and performing edge confidence assessment, blurred edge contours that match the component feature data are selected, improving the reliability of edge recognition under motion conditions. The blurred edge contours are then compared with standard edge contours to analyze the dimensional compliance of the component edge contours. The dimensional compliance rate of all components is statistically analyzed to trigger sorting or line stop operations, thereby improving the automation level and production efficiency of the industrial production line and reducing quality control costs.

Description

Technical Field

[0001] This invention relates to the field of intelligent measurement technology for imaging instruments, and specifically to an intelligent measurement and analysis system for imaging instruments. Background Technology

[0002] In industrial assembly line production, dimensional accuracy inspection of parts is a crucial step in ensuring product quality. Traditional image measurement technology is susceptible to motion blur when dealing with high-speed moving parts, leading to inaccurate edge contour recognition and consequently affecting dimensional measurement accuracy. With the increasing efficiency of assembly line production, higher demands are placed on the speed, accuracy, and intelligence of parts inspection, necessitating an intelligent measurement and analysis system that accurately identifies blurred edges and implements closed-loop feedback processing.

[0003] Existing technologies, such as Chinese Patent Publication No. CN120355656A, disclose a method and system for processing component images based on multi-camera computer vision. This method sets up three 2D cameras based on a preset positional relationship, acquires images of the component based on the 2D cameras, records them as sub-images, and generates a set of sub-images; acquires a set of sub-images of the component in any pose, and groups the sub-image sets; acquires a set of images to be determined, acquires the group F in which the set of images to be determined belongs, and judges the quality of the component based on the group F. This method can efficiently and accurately detect the quality of the component.

[0004] Existing technologies, such as Chinese Patent Publication No. CN118781062A, disclose an automated image detection system. This system divides the target object into several detection zones based on its feature data and light source characteristics. It then determines the optimal illumination parameters for each detection zone based on its material and color characteristics, thereby adjusting the light source during the detection process. This application adjusts the light source according to the feature data of each target object, improving the quality of the acquired image data. This application obtains an illumination feature model that characterizes the mapping relationship between material features, color features, and corresponding optimal illumination parameters, and uses this model to obtain the illumination parameters for each detection zone. Furthermore, this application uses an artificial intelligence model to obtain the illumination parameters for each detection zone, improving the accuracy of light source adjustment and image quality.

[0005] However, existing technologies have the following problems: 1. Existing technologies improve efficiency through continuous motion, but do not dynamically compensate for the blurring caused by the movement of parts. In a production line, when parts are in motion, blurred images can lead to distorted edge contours, directly affecting the accuracy of subsequent part dimension measurements and reducing inspection reliability.

[0006] 2. Existing technologies do not use multi-feature fusion to extract edge contour features. Relying solely on a single image feature makes it difficult to accurately identify and extract blurred edge contours of parts. The accuracy of edge contour recognition under blurry or noisy conditions is low, leading to deviations in the judgment of part edge contour dimensions and reducing the effectiveness of dimension measurement. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the prior art and provide an intelligent measurement and analysis system for imaging instruments, which enables closed-loop control for accurate measurement of the edge contour dimensions of parts in motion scenarios, ensuring high precision and high reliability of dimensional measurement.

[0008] The technical solution adopted by the present invention to solve its technical problem is: an intelligent measurement and analysis system for an image instrument, including an image acquisition module, an image preprocessing module, a fuzzy edge contour recognition module, a fuzzy edge contour measurement module, and a feedback result processing module.

[0009] The connections between the modules are as follows: the image acquisition module is connected to the image preprocessing module; the image preprocessing module is connected to the fuzzy edge contour recognition module; the fuzzy edge contour recognition module is connected to the fuzzy edge contour measurement module; and the fuzzy edge contour measurement module is connected to the feedback result processing module.

[0010] The image acquisition module is used to acquire grayscale images of parts on the industrial production line in real time through an image sensor, and to perform motion blur dynamic compensation on the grayscale images to generate initial part images.

[0011] The image preprocessing module is used to perform adaptive noise suppression and edge enhancement processing on the initial component image to obtain the processed target component image.

[0012] The fuzzy edge contour recognition module is used to extract grayscale gradient data, texture feature data and spatial distribution feature data of target component images based on multi-feature fusion, and to filter out fuzzy edge contours that match the component feature data by performing edge confidence evaluation with the component sample set.

[0013] The fuzzy edge contour measurement module is used to combine the identified fuzzy edge contour with the preset standard edge contour of the component for contour matching and geometric parameter comparison analysis to obtain the dimensional compliance status of the component edge contour.

[0014] The feedback result processing module is used to calculate the dimensional pass rate of all parts on the industrial production line and trigger sorting or line stop operations based on the dimensional pass rate.

[0015] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention acquires grayscale images of parts on the industrial production line in real time through an imager, adjusts the exposure time of the imager based on the blur degree parameter, performs motion blur dynamic compensation on the grayscale image to generate the initial part image, solves the problem of image blur caused by continuous motion, provides a high-quality data foundation for subsequent measurement and analysis, thereby improving the accuracy and stability of size measurement.

[0016] (2) The present invention obtains the processed target component image by performing adaptive noise suppression and edge enhancement processing on the initial component image. By optimizing the image quality in a targeted manner, the edge features of the target component image are made more prominent, thereby improving the accuracy of edge contour recognition and reducing misjudgment caused by image quality.

[0017] (3) Based on multi-feature fusion, the present invention extracts grayscale gradient data, texture feature data and spatial distribution feature data of target component images. By performing edge confidence evaluation with the component sample set, it selects fuzzy edge contours that match the component feature data, improves the reliability of edge recognition under motion conditions, and produces the effect of extracting real fuzzy edge contours with high confidence, ensuring the effectiveness of component size measurement and improving the recognition accuracy of fuzzy edge components.

[0018] (4) This invention compares and analyzes the dimensional compliance of the component edge contours by comparing the fuzzy edge contours with the standard edge contours, calculates the dimensional compliance rate of all components on the industrial production line, triggers sorting or line stop operations based on the dimensional compliance rate, and establishes a closed-loop feedback mechanism to achieve automatic sorting of unqualified components and timely early warning of batch quality problems, thereby improving the automation level and production efficiency of the industrial production line and reducing quality control costs. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the system module connections of the present invention.

[0021] Figure 2 This is a schematic diagram of the process for forming the comprehensive feature matrix in this invention.

[0022] Figure 3 This is a schematic diagram illustrating the specific process of the feedback result processing module in this invention. Detailed Implementation

[0023] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of the invention. Furthermore, it should be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale.

[0024] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the invention or its application or use. Techniques, methods, and apparatus known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and apparatus should be considered part of the specification.

[0025] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0026] Please see Figure 1 As shown, the present invention provides an intelligent measurement and analysis system for an image instrument, including an image acquisition module, an image preprocessing module, a fuzzy edge contour recognition module, a fuzzy edge contour measurement module, and a feedback result processing module.

[0027] The connections between the modules are as follows: the image acquisition module is connected to the image preprocessing module; the image preprocessing module is connected to the fuzzy edge contour recognition module; the fuzzy edge contour recognition module is connected to the fuzzy edge contour measurement module; and the fuzzy edge contour measurement module is connected to the feedback result processing module.

[0028] The image acquisition module is used to acquire grayscale images of parts on the industrial production line in real time through an image sensor, and to perform motion blur dynamic compensation on the grayscale images to generate initial part images.

[0029] It should be noted that the specific content of the image acquisition module is as follows: the position of the parts is monitored in real time by the position sensor on the industrial production line. When the parts reach the preset acquisition area of ​​the imager, the imager takes a picture of the parts that have entered the preset acquisition area, and converts the captured image into a grayscale image through grayscale processing.

[0030] The motion velocity vector and instantaneous displacement of the components are collected in real time by optical sensors. These are then combined with the frame rate parameters of the image capture device to calculate the blurring parameters caused by the motion of the components at the moment of image acquisition.

[0031] The exposure time of the imager is adjusted based on the blur level parameter, and the grayscale image is dynamically compensated for motion blur to generate the initial component image.

[0032] In one specific embodiment, the optical sensor incorporates a miniature laser ranging component and a high-speed image differential analysis component. The miniature laser ranging component emits laser pulses at a set frequency toward the surface of the component and obtains the instantaneous displacement of the component in real time by calculating the pulse reflection time difference. The high-speed image differential analysis component synchronously captures continuous frame images of the component's motion trajectory and obtains the motion velocity vector by analyzing the displacement of adjacent frame pixels, wherein the motion velocity vector includes velocity and direction.

[0033] It should be noted that the calculation component calculates the blurring parameter caused by motion at the moment of image acquisition, specifically as follows: The set shooting frame rate parameter is read from the image sensor's control chip and converted into time-dimensional data to obtain the single-frame acquisition duration. For example, if the shooting frame rate parameter is 30 frames / second, then converting this parameter into time-dimensional data yields the single-frame acquisition duration as follows: Second.

[0034] The motion velocity vector of the component is decomposed into X-axis velocity component and Y-axis velocity component. Combined with the single-frame acquisition duration, the displacement of the component in the X-axis and Y-axis directions during the single-frame acquisition process is calculated.

[0035] The X-axis direction is along the imaging direction of the imager, and the Y-axis direction is perpendicular to the imaging direction. The displacement in the X-axis direction is the X-axis velocity multiplied by the single-frame acquisition time. Similarly, the displacement in the Y-axis direction is the Y-axis velocity multiplied by the single-frame acquisition time.

[0036] If the absolute displacement in the X-axis direction is greater than the absolute displacement in the Y-axis direction, the blur direction is horizontal; if the absolute displacement in the Y-axis direction is greater than the absolute displacement in the X-axis direction, the blur direction is vertical; if the absolute displacement in the Y-axis direction is equal to the absolute displacement in the X-axis direction and is not zero, the blur direction is oblique; and when the absolute displacement in the Y-axis direction and the absolute displacement in the X-axis direction are both zero, there is no blur.

[0037] The blur distance of components during a single frame acquisition is calculated by vector synthesis, and the blur direction and blur distance are integrated into a blur degree parameter.

[0038] The fuzzy distance of the components is ,in This represents the displacement along the X-axis. This represents the displacement along the Y-axis.

[0039] In one specific embodiment, the exposure time adjustment method of the imager is as follows: when the blur distance of the component is greater than the preset maximum acceptable blur distance, the ratio of the preset maximum acceptable blur distance to the blur distance is used as the blur distance correction coefficient; otherwise, 1 is used as the blur distance correction coefficient.

[0040] When the blur direction is horizontal or vertical, it aligns with the pixel arrangement direction of the image sensor, and 1 is used as the blur direction correction coefficient. When the blur direction is diagonal, the sine of the diagonal angle is used as the blur direction correction coefficient. Here, diagonal means aligning with the diagonal direction of the image sensor pixels, and the diagonal angle can be... .

[0041] The adjusted exposure time of the imager is the product of the imager's set baseline exposure time and the blur distance correction coefficient and blur direction correction coefficient.

[0042] This invention acquires grayscale images of parts on an industrial production line in real time using an imager. Based on the blur level parameter, the exposure time of the imager is adjusted, and motion blur is dynamically compensated for to generate initial part images. This solves the problem of image blurring caused by continuous motion, providing a high-quality data foundation for subsequent measurement and analysis, thereby improving the accuracy and stability of dimensional measurement.

[0043] The image preprocessing module performs adaptive noise suppression and edge enhancement processing on the initial component image to obtain the processed target component image. Adaptive noise suppression and edge enhancement processing are existing technologies and will not be described in detail here.

[0044] This invention obtains a processed target component image by performing adaptive noise suppression and edge enhancement processing on the initial component image. By specifically optimizing the image quality, the edge features of the target component image are made more prominent, thereby improving the accuracy of edge contour recognition and reducing misjudgments caused by image quality.

[0045] The fuzzy edge contour recognition module is used to extract grayscale gradient data, texture feature data and spatial distribution feature data of target component images based on multi-feature fusion, and to filter out fuzzy edge contours that match the component feature data by performing edge confidence evaluation with the component sample set.

[0046] like Figure 2 As shown, the multi-feature fusion includes a feature extraction sub-process and a feature fusion sub-process. The feature extraction sub-process extracts the gray-level gradient feature matrix, texture feature vector, and spatial distribution feature set of the target component image respectively. The feature fusion sub-process assigns feature weights to the gray-level gradient feature matrix, texture feature vector, and spatial distribution feature set, and then performs weighted fusion with the corresponding features to form a comprehensive feature matrix.

[0047] Furthermore, the specific content of the feature extraction sub-process is as follows: the target component image is converted to grayscale to obtain a grayscale image of the target component; the Sobel operator is used to perform convolution operation on the grayscale image to obtain the grayscale gradient values ​​of all pixels in the image in the horizontal and vertical directions; the gradient magnitude and gradient direction of each pixel are determined based on the grayscale gradient values ​​of all pixels in the horizontal and vertical directions; and the gradient magnitude and gradient direction of all pixels are integrated into a matrix form to form a grayscale gradient feature matrix.

[0048] The grayscale image of the target component is divided into several sub-regions. Texture feature parameters of each sub-region are extracted based on the gray-level co-occurrence matrix. The texture feature parameters of each sub-region are arranged in a preset order to form a texture feature sub-vector, and then integrated to form a texture feature vector. The texture feature parameters include, but are not limited to, energy, entropy, and contrast.

[0049] Pixels whose grayscale values ​​meet preset edge feature conditions are selected from the grayscale image of the target component and designated as edge points. The coordinates of all edge points are marked by coordinate positioning. Based on the coordinates of all edge points, connected regions are determined, and the geometric attribute data of the connected regions are used to form a spatial distribution feature set. The geometric attribute data of the connected regions includes, but is not limited to, region area, perimeter, and aspect ratio of the circumscribed rectangle.

[0050] In one specific embodiment, the grayscale gradient data is the gradient magnitude and gradient direction of each pixel; the texture feature data is the energy, entropy, and contrast of each sub-region; and the spatial distribution feature data is the area, perimeter, and aspect ratio of the circumscribed rectangle of the connected region.

[0051] The preset edge feature condition can be that the gray value of a pixel has a preset gray value difference threshold compared to the gray values ​​of its neighboring pixels. For example, if the absolute value of the difference between the gray value of a pixel and the gray values ​​of at least 3 pixels in its 8-neighborhood is greater than the set gray value difference threshold, then the pixel is considered an edge point.

[0052] Furthermore, the specific content of the feature fusion sub-process is as follows: the gray-level gradient feature matrix, texture feature vector, and spatial distribution feature set are standardized and dimension-unified to obtain the processed gray-level gradient feature matrix, texture feature matrix, and spatial distribution feature matrix.

[0053] The accuracy rates of gray-level gradient features, texture features, and spatial distribution features in edge recognition are collected from historical sample images. The weights of these features are then determined based on their accuracy rates. Each weight is the ratio of its accuracy rate to the sum of its accuracy rates.

[0054] The weights of gray-level gradient features, texture features, and spatial distribution features are multiplied element-wise with the corresponding feature matrices, and then matrix superposition is performed to form a comprehensive feature matrix.

[0055] The gray-level gradient feature matrix, texture feature vector, and spatial distribution feature set can be standardized using Min-Max standardization. The purpose of standardization is to eliminate weight bias caused by differences in the numerical range of different features, so that all features are within the range of [0, 1].

[0056] The purpose of the dimensional unification process is to align the dimensions of the gray-level gradient feature matrix, texture feature vector, and spatial distribution feature set to facilitate subsequent element-wise multiplication and matrix superposition. Since the gray-level gradient feature matrix is ​​a two-dimensional spatial matrix, the texture feature vector and spatial distribution feature set need to be expanded / converted into two-dimensional spatial matrices of the same dimension. Specifically, the number of pixels in the gray-level gradient feature matrix is ​​used as the number of rows in the two-dimensional spatial matrix, and the maximum number of features in the gray-level gradient feature matrix, texture feature vector, and spatial distribution feature set is used as the number of columns in the two-dimensional spatial matrix. Let's assume... A two-dimensional spatial matrix. For example, the gray-level gradient feature matrix has 2 features, and the texture feature vector and spatial distribution feature set have 3 features. .

[0057] Since the number of columns in the gray-level gradient feature matrix is ​​less than the number of columns in the comprehensive feature matrix, the remaining columns are added to the gray-level gradient feature matrix, and the values ​​of the remaining columns are interpolated using 0 to fill in the gaps, thus forming... The gray-level gradient feature matrix.

[0058] Texture feature vectors are extracted based on sub-regions and need to be mapped to... A two-dimensional spatial matrix is ​​first used to determine the spatial location of each sub-region in the image. Then, the texture feature parameters of all sub-regions are filled into all pixel positions within the corresponding spatial range to form a... The texture feature matrix, where each pixel value is the texture feature parameter of the sub-region.

[0059] Spatial distribution feature sets are extracted based on connected components and need to be mapped to... A two-dimensional spatial matrix is ​​formed by first determining the pixel range of the connected region in the image, then filling the corresponding pixel positions within the region with the geometric attribute data of the connected region, and filling all pixels outside the region with 0, ultimately forming... The spatial distribution characteristic matrix.

[0060] It should be noted that the method for selecting the fuzzy edge contours that conform to the component feature data is as follows: extract the edge contour images of different types of components with different degrees of fuzziness and their corresponding real edge contour feature information from the component sample set, compare the similarity of the resulting comprehensive feature matrix with the real edge contour feature information, output the confidence value of the edge contour based on the maximum matching similarity, and extract the type of component with that degree of fuzziness when the confidence value is greater than the preset confidence threshold, and use its edge contour as the fuzzy edge contour.

[0061] The matching similarity comparison method is to use cosine similarity to calculate the matching similarity between the comprehensive feature matrix and the real edge contour feature information converted into a standardized edge contour feature matrix.

[0062] The confidence value of the edge contour is output as follows: taking the maximum matching similarity as input, the confidence value of the edge contour is output through the sigmoid activation function. The sigmoid activation function can non-linearly map the matching similarity to a confidence value in the interval [0, 1], and the higher the matching similarity, the closer the confidence value is to 1.

[0063] This invention extracts grayscale gradient data, texture feature data, and spatial distribution feature data from target component images based on multi-feature fusion. By performing edge confidence assessment with a component sample set, it filters out fuzzy edge contours that match the component feature data, improving the reliability of edge recognition under motion conditions. This results in the extraction of real fuzzy edge contours with high confidence, ensuring the effectiveness of component size measurement and improving the accuracy of identifying fuzzy edge components.

[0064] The fuzzy edge contour measurement module is used to combine the identified fuzzy edge contour with the preset standard edge contour of the component for contour matching and geometric parameter comparison analysis to obtain the dimensional compliance status of the component edge contour.

[0065] It should be noted that the contour matching and geometric parameter comparison analysis process is as follows: the fuzzy edge contour is matched with the standard edge contour of the corresponding type of component to obtain the overall shape of the contour; the contour areas that deviate from the standard edge contour are marked as suspicious areas, and the area ratio of the suspicious areas is obtained; the feature position comparison is performed on the remaining areas other than the suspicious areas, the position deviation distance between the fuzzy edge contour and the standard edge contour at key feature points is calculated, and the area ratio of the suspicious areas and the position deviation distance are weighted and fused to obtain the contour matching deviation degree.

[0066] Key feature points can include corner points, endpoints, and arc vertices.

[0067] Geometric parameter data is extracted from the blurred edge contour and compared with the corresponding geometric parameter data in the preset standard edge contour of the component to obtain the geometric parameter deviation. The geometric parameter data includes, but is not limited to, line segment length, the angle between adjacent line segments, and the radius of the arc contour.

[0068] In one specific embodiment, the weighting method for the suspicious area proportion and the positional deviation distance is as follows: multiple sets of known qualified and unqualified component edge contour samples are collected. For each sample, the suspicious area proportion, the positional deviation distance of key feature points, and the actual quality judgment result are recorded, where qualified quality is set to 1 and unqualified quality is set to 0. The suspicious area proportion and the positional deviation distance of key feature points are Z-score standardized. The processed suspicious area proportion and the positional deviation distance of key feature points are substituted into the established logistic regression equation. The least squares method is used to solve the optimal regression coefficient set in the logistic regression equation. The ratio of the regression coefficient corresponding to the suspicious area proportion to the sum of the optimal regression coefficient set is used as the weight of the suspicious area proportion, and the ratio of the regression coefficient corresponding to the positional deviation distance to the sum of the optimal regression coefficient set is used as the weight of the positional deviation distance.

[0069] The geometric parameter deviation is the average of the ratios of the absolute differences of all geometric parameter data to the corresponding geometric parameter data in the standard edge contour of the component.

[0070] It should be noted that the method for analyzing the dimensional compliance of the component edge contour is as follows: if the contour matching deviation of the component is greater than the set contour matching deviation threshold or the geometric parameter deviation is greater than the set geometric parameter deviation threshold, then the dimensional compliance of the component edge contour is unqualified; otherwise, the dimensional compliance of the component edge contour is qualified.

[0071] The feedback result processing module is used to calculate the dimensional pass rate of all parts on the industrial production line and trigger sorting or line stop operations based on the dimensional pass rate.

[0072] like Figure 3 As shown, the specific method of the feedback result processing module is as follows: obtain the edge contour dimension qualification status of all parts on the industrial production line, count the number of parts with qualified edge contour dimensions, and use the ratio of the qualified edge contour dimensions to the total number of parts as the dimension qualification rate.

[0073] If the dimensional pass rate is less than the set dimensional pass rate threshold, the industrial production line will be stopped. Otherwise, parts marked as having unqualified edge contour dimensions will be sorted.

[0074] This invention analyzes the dimensional compliance of component edge contours by comparing and contrasting fuzzy edge contours with standard edge contours, calculates the dimensional compliance rate of all components on the industrial production line, and triggers sorting or line stop operations based on the dimensional compliance rate. By establishing a closed-loop feedback mechanism, it achieves automatic sorting of non-conforming components and timely early warning of batch quality problems, thereby improving the automation level and production efficiency of industrial production lines and reducing quality control costs.

[0075] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0076] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0077] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0078] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0079] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0080] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An intelligent measurement and analysis system for an image analyzer, characterized in that, include: The image acquisition module is used to acquire grayscale images of parts on the industrial production line in real time through an image sensor, and to perform motion blur dynamic compensation on the grayscale images to generate initial part images. The image preprocessing module is used to perform adaptive noise suppression and edge enhancement processing on the initial component image to obtain the processed target component image. The fuzzy edge contour recognition module is used to extract grayscale gradient data, texture feature data and spatial distribution feature data of target component images based on multi-feature fusion, and to filter out fuzzy edge contours that match the component feature data by performing edge confidence evaluation with the component sample set. The fuzzy edge contour measurement module is used to combine the identified fuzzy edge contour with the preset standard edge contour of the parts for contour matching and geometric parameter comparison analysis to obtain the dimensional compliance status of the edge contour of the parts. The feedback result processing module is used to calculate the dimensional pass rate of all parts on the industrial production line and trigger sorting or line stop operations based on the dimensional pass rate.

2. The intelligent measurement and analysis system for an image analyzer according to claim 1, characterized in that: The specific contents of the image acquisition module are as follows: The position of parts is monitored in real time by position sensors on the industrial production line. When a part reaches the preset acquisition area of ​​the imager, the imager takes a picture of the part that has entered the preset acquisition area and converts the captured image into a grayscale image through grayscale processing. The motion velocity vector and instantaneous displacement of the components are collected in real time by optical sensors. These are then combined with the frame rate parameters of the image capture device to calculate the blurring parameters of the components caused by motion at the moment of image acquisition. The exposure time of the imager is adjusted based on the blur level parameter, and the grayscale image is dynamically compensated for motion blur to generate the initial component image.

3. The intelligent measurement and analysis system for an image analyzer according to claim 2, characterized in that: The parameters describing the degree of blurring caused by motion of the calculated components at the moment of image acquisition are as follows: The set shooting frame rate parameter is read from the control chip of the image device and converted into time dimension data to obtain the single frame acquisition duration; The motion velocity vector of the component is decomposed into X-axis velocity component and Y-axis velocity component. Combined with the single frame acquisition time, the displacement of the component in the X-axis and Y-axis directions during the single frame acquisition process is calculated. If the absolute displacement in the X-axis direction is greater than the absolute displacement in the Y-axis direction, the fuzzy direction is horizontal; if the absolute displacement in the Y-axis direction is greater than the absolute displacement in the X-axis direction, the fuzzy direction is vertical; if the absolute displacement in the Y-axis direction is equal to the absolute displacement in the X-axis direction and is not zero, the fuzzy direction is oblique. The blur distance of components during a single frame acquisition is calculated by vector synthesis, and the blur direction and blur distance are integrated into a blur degree parameter.

4. The intelligent measurement and analysis system for an image analyzer according to claim 1, characterized in that: The multi-feature fusion includes a feature extraction sub-process and a feature fusion sub-process. The feature extraction sub-process extracts the gray-level gradient feature matrix, texture feature vector, and spatial distribution feature set of the target component image, respectively. The feature fusion sub-process assigns feature weights to the gray-level gradient feature matrix, texture feature vector, and spatial distribution feature set, and then performs weighted fusion with the corresponding features to form a comprehensive feature matrix.

5. The intelligent measurement and analysis system for an image analyzer according to claim 4, characterized in that: The specific details of the feature extraction sub-process are as follows: The target component image is converted to grayscale to obtain a grayscale image of the target component. The grayscale gradient values ​​of all pixels in the image are obtained in the horizontal and vertical directions. The gradient magnitude and gradient direction of each pixel are determined based on the grayscale gradient values ​​of all pixels in the horizontal and vertical directions. The gradient magnitude and gradient direction of all pixels are integrated into a matrix form to form a grayscale gradient feature matrix. The grayscale image of the target component is divided into several sub-regions. The texture feature parameters of each sub-region are extracted. The texture feature parameters of each sub-region are arranged in a preset order to form a texture feature sub-vector. The sub-vectors are then integrated to form a texture feature vector. Pixels whose grayscale values ​​meet the preset edge feature conditions are selected from the grayscale image of the target component. The coordinates of all selected pixels are obtained by coordinate positioning and used as candidate feature point coordinates. Cluster analysis is performed on the candidate feature point coordinates to form a spatial distribution feature set.

6. The intelligent measurement and analysis system for an image analyzer according to claim 4, characterized in that: The specific details of the feature fusion sub-process are as follows: The gray-level gradient feature matrix, texture feature vector, and spatial distribution feature set are standardized and dimension-unified to obtain the processed gray-level gradient feature matrix, texture feature matrix, and spatial distribution feature matrix. Collect the accuracy of gray-level gradient features, texture features and spatial distribution features in edge recognition from historical sample images, and determine the proportion and weight of gray-level gradient features, texture features and spatial distribution features based on the accuracy. The weights of gray-level gradient features, texture features, and spatial distribution features are multiplied element-wise with the corresponding feature matrices, and then matrix superposition is performed to form a comprehensive feature matrix.

7. The intelligent measurement and analysis system for an image analyzer according to claim 6, characterized in that: The method for filtering the fuzzy edge contours that match the component feature data is as follows: The edge contour images of different types of parts under different degrees of fuzziness and their corresponding real edge contour feature information are extracted from the part sample set. The resulting comprehensive feature matrix is ​​matched and compared with the real edge contour feature information. The confidence value of the edge contour is output based on the maximum matching similarity. When the confidence value is greater than the preset confidence threshold, the type of part under that degree of fuzziness is extracted and its edge contour is used as the fuzzy edge contour.

8. The intelligent measurement and analysis system for an image analyzer according to claim 7, characterized in that: The process of contour matching and geometric parameter comparison analysis is as follows: The blurry edge contour is matched with the standard edge contour of the corresponding type of component to obtain the overall shape of the contour. The contour areas that deviate from the standard edge contour are marked as suspicious areas and the area ratio of the suspicious areas is obtained. The feature position comparison is performed on the remaining areas other than the suspicious areas to calculate the position deviation distance between the blurry edge contour and the standard edge contour at key feature points. The area ratio of the suspicious areas and the position deviation distance are weighted and fused to obtain the contour matching deviation degree. Geometric parameter data is extracted from the fuzzy edge contour, and the absolute difference is compared with the corresponding geometric parameter data in the preset standard edge contour of the component to obtain the geometric parameter deviation.

9. The intelligent measurement and analysis system for an image analyzer according to claim 8, characterized in that: The method for analyzing the dimensional compliance of the component edge contour is as follows: If the deviation of the contour matching of a component is greater than the set threshold for contour matching or the deviation of the geometric parameters is greater than the set threshold for geometric parameters, then the dimensional compliance of the component's edge contour is unqualified; otherwise, the dimensional compliance of the component's edge contour is qualified.

10. The intelligent measurement and analysis system for an image analyzer according to claim 1, characterized in that: The specific method of the feedback result processing module is as follows: Obtain the edge contour dimension compliance status of all parts on the industrial production line, count the number of parts with qualified edge contour dimensions, and use the ratio of the qualified edge contour dimensions to the total number of parts as the dimension compliance rate. If the dimensional pass rate is less than the set dimensional pass rate threshold, the industrial production line will be stopped. Otherwise, parts marked as having unqualified edge contour dimensions will be sorted.

Images