Blade shutter-based visual inspection method and system

Abstract

The application relates to a blade shutter-based visual detection method and system, relates to the technical field of computer vision and image processing, and comprises the following steps: collecting a blade shutter detection image and a historical deformation degree of a blade shutter; analyzing the blade shutter detection image to determine the deformation degree of the blade shutter; judging whether the deformation degree of the blade shutter meets the requirements of the historical deformation degree; if yes, collecting a historical position-state mapping relationship and determining the historical position-state mapping relationship as an actual position-state mapping relationship; if not, adjusting a preset standard position-state mapping relationship according to the deformation degree of the blade shutter to generate the actual position-state mapping relationship; collecting feature point positions of the blade shutter; and determining a detection state of the blade shutter according to the feature point positions and the actual position-state mapping relationship. The application has the effect of improving the detection accuracy of the blade shutter.

Description

Technical Field

[0001] This application relates to the technical field of computer vision and image processing, and in particular to a visual inspection method and system based on a knife switch. Background Technology

[0002] A disconnector is an important high-voltage switching device in a power system. It is mainly used to disconnect or close circuits under no-load or very low-current conditions to achieve the isolation, maintenance and safe operation of electrical equipment.

[0003] In related technologies, the detection of the open or closed status of a disconnector is usually based on visual recognition. The location of feature points on the disconnector is determined by image acquisition and recognition. Then, the mapping relationship between the feature point location and the disconnector status is used. For example, when the disconnector is in the closed state, the feature point location should be within a certain range, while when the disconnector is in the open state, the feature point location should be within another range, thereby determining the current status of the disconnector.

[0004] Regarding the aforementioned technologies, when the disconnector deforms, the position of the feature points on the disconnector changes. When the state of the disconnector changes, the position of the feature points may still be within the position range corresponding to the previous state, which may cause misjudgment of the disconnector state, resulting in low detection accuracy of the disconnector. There is still room for improvement. Summary of the Invention

[0005] To improve the detection accuracy of disconnect switches, this application provides a visual inspection method and system based on disconnect switches.

[0006] Firstly, this application provides a visual inspection method based on a knife switch, employing the following technical solution:

[0007] Visual inspection methods based on disconnect switches include:

[0008] Collect images of the disconnect switch and its historical deformation.

[0009] The detection images of the disconnect switch are analyzed to determine the degree of deformation of the disconnect switch;

[0010] Determine whether the degree of deformation of the disconnect switch meets the requirements of the historical deformation degree;

[0011] If the conditions are met, the historical location status mapping relationship is collected, and the historical location status mapping relationship is determined as the actual location status mapping relationship;

[0012] If it does not meet the requirements, the preset standard position state mapping relationship will be adjusted according to the degree of deformation of the switch to generate the actual position state mapping relationship.

[0013] Collect the location of feature points of the disconnect switch;

[0014] The detection status of the disconnector is determined based on the mapping relationship between the feature point position and the actual position status.

[0015] Optionally, the step of analyzing the disconnector detection image to determine the degree of disconnector deformation includes:

[0016] Extract contour feature points from the knife switch detection image to generate a set of contour feature points;

[0017] The bending deformation of the knife gate is generated by calculating based on the set of contour feature points and the preset standard contour line.

[0018] The torsional deformation of the knife gate is calculated based on the set of contour feature points and the preset set of standard contour points.

[0019] The degree of expansion and contraction of the knife switch is calculated based on the set of contour feature points and the preset standard feature point spacing.

[0020] The bending deformation, torsional deformation, and expansion deformation of the disconnector are analyzed based on a pre-defined disconnector deformation rule library to determine the degree of disconnector deformation.

[0021] Optionally, the step of calculating the bending deformation of the knife gate based on the set of contour feature points and a preset standard contour line includes:

[0022] Iterate through the Euclidean distances between the set of contour feature points and the standard contour line to generate the line distances between the contour feature points;

[0023] The straight-line distance of the contour feature points is normalized based on the preset maximum straight-line distance to generate the curvature of the contour feature points.

[0024] The mean value of the curvature of the contour feature points is calculated to generate the bending deformation degree of the knife gate.

[0025] Optionally, the step of calculating the torsional deformation of the knife gate based on the set of contour feature points and a preset set of standard contour points includes:

[0026] Extract the positions of the connected feature points and the standard feature points from the contour feature point set and the standard contour point set, respectively.

[0027] Determine the connection feature point vector based on the location of the connection feature points;

[0028] Determine the standard feature point vector based on the position of the standard feature point;

[0029] Calculate the angle between the connecting feature point vector and the standard feature point vector to generate the torsional deformation of the switch.

[0030] Optionally, the step of calculating the telescopic deformation degree of the knife gate based on the set of contour feature points and the preset standard feature point spacing includes:

[0031] Extract the position of the feature points at the end of the knife switch from the set of contour feature points;

[0032] Calculate the distance between the feature points at the end of the disconnector to generate the fixed length of the disconnector;

[0033] Calculate the difference between the fixed length of the disconnect switch and the standard feature point spacing to generate the disconnect switch extension distance;

[0034] Calculate the quotient of the disconnector's telescopic distance and the standard feature point spacing to generate the telescopic deformation degree of the disconnector.

[0035] Optionally, the step of adjusting the preset standard position state mapping relationship according to the degree of deformation of the switch to generate the actual position state mapping relationship includes:

[0036] Based on the degree of deformation of the disconnect switch, find the corresponding disconnect switch deformation influence coefficient in the preset deformation influence relationship;

[0037] The bending deformation, torsional deformation, and expansion deformation of the disconnect switch are normalized and summed to generate the deformation influence value.

[0038] Collect and adjust the direction coefficient;

[0039] Calculate the product between the adjustment direction coefficient, the knife switch deformation influence coefficient, and the deformation influence value to generate the position adjustment parameters;

[0040] The standard position state mapping relationship is adjusted according to the position adjustment parameters to generate the actual position state mapping relationship.

[0041] Optionally, the steps for acquiring the feature point locations of the disconnector include:

[0042] The gradient of pixels in the knife switch detection image is calculated according to the preset local window parameters to generate the horizontal gradient and vertical gradient of the feature points.

[0043] Construct the gradient matrix of the feature points based on their horizontal and vertical gradients.

[0044] The gradient matrix of the feature points is calculated based on the preset response function to generate the feature point response values;

[0045] The feature point response values ​​are filtered and sorted according to a preset response threshold to determine the maximum response value;

[0046] Collect the initial position of the feature point corresponding to the maximum response value;

[0047] The initial positions of the feature points are matched to generate the feature point positions.

[0048] Optionally, the step of matching the initial positions of feature points to generate feature point positions includes:

[0049] The feature point neighborhood image is extracted from the knife switch detection image based on the initial position of the feature point and the preset neighborhood image parameters.

[0050] The feature point pixels and the average neighborhood pixels are determined based on the feature point neighborhood image.

[0051] Calculate the similarity between feature point pixels, the average neighboring pixels, and preset standard image pixels, and filter out the maximum value to generate feature point matching degree;

[0052] Collect the feature point matching position corresponding to the feature point matching degree;

[0053] The position fusion weights are determined based on the feature point matching degree.

[0054] The feature point matching position and the initial position of the feature point are fused according to the position fusion weight to generate the feature point position.

[0055] Secondly, this application provides a visual inspection system for a knife switch, which adopts the following technical solution:

[0056] A visual inspection system for a knife switch includes:

[0057] The acquisition module is used to acquire images of the disconnect switch, historical deformation levels, historical positional state mapping relationships, and feature point locations.

[0058] A memory for storing the program of the knife switch-based visual inspection method as described in any of the above;

[0059] The processor and the program in the memory can be loaded and executed by the processor to implement the knife switch-based visual inspection method as described in any of the above.

[0060] In summary, this application includes at least one of the following beneficial technical effects:

[0061] 1. By collecting and analyzing the detection images of the disconnect switch, when it is determined that the degree of deformation of the disconnect switch does not meet the requirements of the historical deformation degree, the standard position state mapping relationship is adjusted according to the degree of deformation of the disconnect switch, so as to obtain the actual position state mapping relationship that conforms to the current form of the disconnect switch, ensuring the accuracy of the correspondence between the position of the disconnect switch feature point and the position of the disconnect switch, thereby improving the accuracy of the disconnect switch state detection.

[0062] 2. By analyzing the bending deformation degree, torque deformation degree, and telescopic deformation degree of the disconnector through the disconnector deformation rule library, the degree of disconnector deformation is obtained. The deformation type that affects the position of feature points is included in the analysis scope of the degree of deformation to ensure the accuracy and high correlation of the degree of disconnector deformation.

[0063] 3. By finding the corresponding deformation influence coefficient of the disconnector in the deformation influence relationship based on the degree of disconnector deformation, the bending deformation degree, torsional deformation degree, and telescopic deformation degree of the disconnector are normalized and summed to obtain the deformation influence value. Then, the product between the adjustment direction coefficient, the disconnector deformation influence coefficient, and the deformation influence value is calculated to obtain the position adjustment parameter. This ensures that the position adjustment increases with the degree of deformation and that the adjustment value of the specific position changes with the degree of deformation, thereby improving the accuracy of the mapping relationship of the actual position state. Attached Figure Description

[0064] Figure 1 This is a flowchart of the visual inspection method based on the knife switch in the embodiments of this application.

[0065] Figure 2 This is a flowchart of the steps in this application embodiment to analyze the detection image of the disconnect switch to determine the degree of deformation of the disconnect switch.

[0066] Figure 3 This is a flowchart of the steps in this application embodiment to calculate the bending deformation of the knife gate based on a set of contour feature points and a preset standard contour line.

[0067] Figure 4 This is a flowchart of the steps in this application embodiment to calculate the torsional deformation of the knife gate based on the set of contour feature points and the preset set of standard contour points.

[0068] Figure 5 This is a flowchart of the steps in this application embodiment to calculate the extension and contraction degree of the knife switch based on the set of contour feature points and the preset standard feature point spacing.

[0069] Figure 6 This is a flowchart of the steps in this application embodiment to adjust the preset standard position state mapping relationship according to the degree of deformation of the knife gate in order to generate the actual position state mapping relationship.

[0070] Figure 7 This is a flowchart of the steps for collecting the feature point positions of the disconnector in the embodiments of this application.

[0071] Figure 8 This is a flowchart of the steps in this application embodiment to match the initial position of feature points in order to generate the position of feature points. Detailed Implementation

[0072] To make the purpose, technical solution, and advantages of this application clearer, the following description is provided in conjunction with the appendix. Figures 1 to 8 The present application will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the application.

[0073] Reference Figure 1 This application discloses a visual inspection method based on a knife switch, including the following steps:

[0074] Step S100: Collect the switch detection image and historical deformation degree of the switch.

[0075] Among them, the disconnect switch detection image refers to the image captured by the disconnect switch. A high-definition industrial camera is installed in the horizontal plane of the disconnect switch, and the disconnect switch is continuously photographed by the high-definition industrial camera to obtain the disconnect switch detection image. The disconnect switch detection image is sent to the processing terminal. On the one hand, the disconnect switch detection image can be used to identify the morphological features of the disconnect switch to determine whether the disconnect switch has been deformed. On the other hand, the disconnect switch detection image can be used to determine the position of feature points on the disconnect switch, and the current state of the disconnect switch can be determined according to the mapping relationship between the position of feature points and the open or closed state of the disconnect switch.

[0076] The historical deformation degree refers to the degree of deformation of the disconnector detected at the previous moment. It is obtained by the processing terminal through analysis and calculation of feature parameters related to the disconnector's shape in historical disconnector images. The specific analysis method is the same as the method for determining the degree of disconnector deformation in step S101, and will not be elaborated here. By determining the historical deformation degree, comparative data is provided for subsequently determining whether the disconnector's deformation has changed again, thereby determining whether it is necessary to adjust the mapping relationship between the disconnector's feature point positions and the disconnector's state.

[0077] Step S101: Analyze the disconnector detection image to determine the degree of disconnector deformation.

[0078] The degree of deformation of the disconnector refers to the extent of deformation. In this embodiment, the degree of deformation of the disconnector is described by three levels: Level 1, Level 2, and Level 3. Level 1 deformation represents slight deformation of the disconnector, requiring slight adjustment to the mapping relationship between the disconnector's feature point position and its state. Level 2 deformation represents moderate deformation, requiring a greater adjustment than Level 1. Level 3 deformation is severe, causing the disconnector to fail to complete the normal closing action and triggering an alarm. The degree of deformation is obtained by the processing terminal through feature extraction and analysis of the disconnector detection image. The specific method is described in [reference needed]. Figure 2 The steps.

[0079] Step S102: Determine whether the degree of deformation of the switch meets the requirements of the historical deformation degree.

[0080] The requirement for the degree of historical deformation refers to being consistent with the degree of historical deformation.

[0081] By processing the terminal to determine whether the degree of deformation of the disconnector is consistent with the historical degree of deformation, it can be determined whether a new situation has occurred in the deformation of the disconnector, thus providing data support for whether it is necessary to adjust the mapping relationship between the position of the disconnector feature points and the disconnector state in the future.

[0082] Step S1021: If the conditions are met, collect the historical location status mapping relationship and determine the historical location status mapping relationship as the actual location status mapping relationship.

[0083] If the processing terminal determines that the degree of deformation of the disconnect switch is consistent with the historical degree of deformation, it indicates that there is no new situation in the deformation of the disconnect switch. Therefore, it is not necessary to adjust the mapping relationship between the position of the disconnect switch feature point and the disconnect switch state. Instead, the historical position and state mapping relationship is called and determined as the actual position and state mapping relationship to ensure the accuracy of disconnect switch state detection.

[0084] The historical position and state mapping relationship refers to the mapping relationship between the position of the disconnector feature point and the disconnector state at the previous moment. Taking the horizontal coordinate position as an example, the disconnector feature point position is obtained by the processing terminal by adjusting the mapping relationship between the standard disconnector feature point position and the disconnector state according to the historical deformation degree. The processing terminal backs up the data for later use. The specific adjustment method is the same as the method of adjusting the standard position and state mapping relationship according to the disconnector deformation degree in step S1022, and will not be described in detail here.

[0085] The actual position and state mapping relationship refers to the mapping relationship between the current position of the disconnector feature point and the disconnector state. The actual position and state mapping relationship in this step is the historical position and state mapping relationship.

[0086] Step S1022: If it does not meet the requirements, the preset standard position state mapping relationship is adjusted according to the degree of deformation of the switch to generate the actual position state mapping relationship.

[0087] If the processing terminal determines that the degree of deformation of the disconnector is inconsistent with the historical degree of deformation, it indicates that a new situation has occurred in the deformation of the disconnector. Therefore, it is necessary to adjust the mapping relationship between the location of the disconnector feature point and the disconnector state. In this way, the standard position state mapping relationship is adjusted according to the degree of disconnector deformation to obtain the actual position state mapping relationship, so as to ensure the accuracy of disconnector state detection.

[0088] The standard position and state mapping relationship refers to the mapping relationship between the position of the characteristic point of the disconnector without any deformation and the state of the disconnector. The operator collects images of the disconnector when it is open and closed without deformation, and identifies the position of the characteristic point in the image, thereby forming a mapping table that corresponds the position of the characteristic point to the state of the disconnector.

[0089] The actual position state mapping relationship in this step is consistent with that in step S1021. The difference is that the actual position state mapping relationship in this step is obtained by the processing terminal adjusting the preset standard position state mapping relationship according to the degree of deformation of the switch. For the specific method, refer to [link to relevant documentation]. Figure 6 The steps.

[0090] Step S103: Collect the location of the feature points of the disconnect switch.

[0091] The feature point location refers to the coordinates of the feature points on the disconnect switch. For specific acquisition methods, please refer to [reference needed]. Figure 7 The steps involve determining the location of the characteristic points of the disconnector to provide data support for determining the current state of the disconnector.

[0092] Step S104: Determine the switch detection status based on the mapping relationship between feature point positions and actual position states.

[0093] Among them, the disconnector detection status refers to the detection result of the disconnector status, including the disconnector closed status and the disconnector open status, which is obtained by the processing terminal by looking up the corresponding mapping table of the actual position status mapping relationship according to the horizontal coordinate of the feature point position.

[0094] Reference Figure 2 The steps for analyzing the disconnect switch inspection images to determine the degree of disconnect switch deformation include:

[0095] Step S200: Extract contour feature points from the knife switch detection image to generate a set of contour feature points.

[0096] The contour feature point set refers to the set of pixel coordinates of the knife switch contour. First, the knife switch detection image is preprocessed, including Gaussian filtering for noise reduction, histogram equalization for contrast enhancement, and adaptive thresholding for binarization. Then, the Canny edge detection operator is used to extract edge features from the preprocessed image to determine the closed contour of the knife switch. Finally, corner detection, curvature extremum point method, or key structure point annotation method are used to extract feature points on the closed contour of the knife switch, thus assembling the coordinates of the extracted feature points into an ordered set, which is the contour feature point set. Determining the contour feature point set provides analytical data for subsequent analysis of the knife switch's deformation degree.

[0097] Step S201: Calculate the bending deformation degree of the knife gate based on the set of contour feature points and the preset standard contour line.

[0098] After determining the set of contour feature points, the processing terminal calculates the bending deformation degree of the knife gate based on the coordinates of the feature points in the set and the standard contour line. The specific method is described in [reference needed]. Figure 3 These steps provide data support for subsequently determining the degree of deformation of the disconnector.

[0099] The standard profile straight line refers to the profile straight line equation of the disconnector without deformation, which is obtained by the processing terminal by fitting the coordinates of the feature points at both ends of the disconnector.

[0100] The bending deformation of the knife gate is an index that reflects the degree of bending of the knife gate. It is obtained by the processing terminal calculating the difference between the distance and the maximum distance between the coordinates of different feature points in the set of contour feature points and the standard contour line. The bending deformation of the knife gate is in the range of 0 to 1. The larger the bending deformation of the knife gate, the more severe the bending of the knife gate. After the feature points are offset along the bending direction, the mapping relationship between the standard position and state is linearly offset, and translation along the bending direction is required.

[0101] Step S202: Calculate the torsional deformation degree of the knife gate based on the set of contour feature points and the preset set of standard contour points.

[0102] In this process, after determining the set of contour feature points, the processing terminal generates a feature point vector based on the coordinates of the feature points in the set. It then compares the angle between this feature point vector and the feature point vector formed by the standard contour point set to obtain the torsional deformation degree of the switch. The specific method is described in [reference needed]. Figure 4 These steps provide data support for subsequently determining the degree of deformation of the disconnector.

[0103] The standard contour point set refers to the set of contour feature point coordinates of the undeformed knife switch. It is obtained by the processing terminal processing the undeformed knife switch image. The specific method is the same as the extraction method of the contour feature point set in step S200, and will not be described in detail here.

[0104] The torsional deformation degree of the disconnector is an index reflecting the degree of torsion of the disconnector. It is obtained by the processing terminal forming a feature point vector based on the coordinates of feature points in the contour feature point set, and comparing the vector angle between the feature point vector and the feature point vector formed by the standard contour point set. The range of the torsional deformation degree of the disconnector is from 0 to π / 2. The larger the torsional deformation degree of the disconnector, the more severe the torsion of the disconnector. After torsion occurs, the position of the feature point shifts to the torsional side, and the offset needs to be corrected to ensure the accurate correspondence between the position and state after torsion.

[0105] Step S203: Calculate the extension and contraction degree of the knife gate based on the set of contour feature points and the preset standard feature point spacing.

[0106] After determining the set of contour feature points, the processing terminal determines the degree of expansion and contraction of the disconnector based on the difference between the distance between the feature points at both ends of the disconnector and the standard feature point spacing. The specific method is described in [reference needed]. Figure 5 This process provides data support for subsequently determining the degree of deformation of the switch.

[0107] The standard feature point spacing refers to the distance between the feature points at both ends of the disconnector that has not undergone deformation, which is obtained by actual measurement by the operator.

[0108] The degree of deformation of the disconnect switch refers to an index reflecting the extent of deformation of the disconnect switch. It is determined by the processing terminal based on the difference between the distance between the feature points at both ends of the disconnect switch in the contour feature point set and the distance between the standard feature points. The range of the degree of deformation of the disconnect switch is 0 to 0.15. If it exceeds 0.15, it is considered severe deformation and is not included in the analysis range. The larger the degree of deformation of the disconnect switch, the more severe the deformation of the disconnect switch. After deformation occurs, the feature points on the disconnect switch are translated as a whole. At this time, it is necessary to translate the correspondence between the standard position and the state to adapt to the deformation of the disconnect switch.

[0109] Step S204: Analyze the bending deformation degree, torsional deformation degree, and expansion deformation degree of the disconnector according to the preset disconnector deformation rule library to determine the degree of disconnector deformation.

[0110] In this process, after determining the bending deformation, torsional deformation, and expansion deformation of the disconnect switch, the processing terminal analyzes these parameters according to the disconnect switch deformation rule library to determine the degree of disconnect switch deformation, thereby ensuring the accuracy of the degree of disconnect switch deformation.

[0111] The disconnect switch deformation rule library refers to the rules for determining the deformation level of the disconnect switch based on three-dimensional data. The determination rule for Level 1 deformation is that the bending deformation of the disconnect switch is less than 0.1, the torsional deformation of the disconnect switch is less than π / 18, and the expansion and contraction deformation of the disconnect switch is less than 0.05. The determination rule for Level 2 deformation is that the bending deformation of the disconnect switch is greater than 0.1 and less than 0.3, or the torsional deformation of the disconnect switch is greater than π / 18 and less than π / 9, or the expansion and contraction deformation of the disconnect switch is greater than 0.05 and less than 0.15. The determination rule for Level 3 deformation is that the bending deformation of the disconnect switch is greater than 0.3, or the torsional deformation of the disconnect switch is greater than π / 9, or the expansion and contraction deformation of the disconnect switch is greater than 0.15.

[0112] Reference Figure 3 The steps for generating the bending deformation of the knife gate based on the set of contour feature points and the preset standard contour line include:

[0113] Step S300: Traverse the Euclidean distance between the set of contour feature points and the standard contour line to generate the straight-line distance between the contour feature points.

[0114] The straight-line distance between contour feature points refers to the shortest distance between all feature points in the contour feature point set and the standard contour line. The processing terminal substitutes the coordinates of all feature points in the contour feature point set and the equation of the corresponding line of the standard contour line into the point-to-line distance formula for calculation, thus obtaining the distance from different feature points to the standard contour line. By determining the straight-line distance between contour feature points, the degree to which the contour feature points deviate from the contour line can be determined, providing data support for subsequent determination of contour curvature.

[0115] Step S301: Normalize the straight-line distance of the contour feature points according to the preset maximum straight-line distance to generate the curvature of the contour feature points.

[0116] Among them, the maximum distance between the feature points of the contour line refers to the maximum distance between the feature points in the contour of the disconnector and the standard contour line when the disconnector has not deformed. It is obtained by the processing terminal after analyzing and calculating the standard disconnector image using the same method as in step S300.

[0117] The curvature of the contour feature point refers to the degree of deviation between the current detected feature point and the straight line distance and the standard distance. It is obtained by the processing terminal calculating the quotient between the straight line distance of the contour feature point and the maximum straight line distance of the contour feature point. The greater the curvature of the contour feature point, the more serious the offset of the contour feature point.

[0118] Step S302: Calculate the mean value of the curvature of the contour feature points to generate the bending deformation degree of the knife gate.

[0119] In this process, after determining the curvature of all feature points in the outline of the disconnector, the mean value of the curvature of the feature points is calculated to obtain the bending deformation of the disconnector, ensuring the accuracy of the bending deformation and avoiding the randomness of the bending deformation.

[0120] Reference Figure 4 The steps for generating the torsional deformation of the knife gate based on the set of contour feature points and the preset set of standard contour points include:

[0121] Step S400: Extract the positions of the connected feature points and the standard feature points from the contour feature point set and the standard contour point set, respectively.

[0122] Among them, the connection feature point position refers to the coordinates of the feature points of the current connection part on the disconnector, such as the disconnector endpoint and the disconnector hinge point, which are obtained by the processing terminal from the set of contour feature points.

[0123] The standard feature point position refers to the coordinates of the feature points on the connection parts of the standard disconnector, such as the disconnector endpoint and the disconnector hinge point, which are obtained by the processing terminal from the standard contour point set.

[0124] By determining the positions of the connecting feature points and the standard feature points, when the switch twists, the positions of the feature points change, which causes the vectors between the feature points in the connecting parts to change. The degree of twist can be determined by the angle between the vectors before and after the change.

[0125] Step S401: Determine the connecting feature point vector based on the location of the connecting feature points.

[0126] The connection feature point vector refers to the vector between the current disconnector connection points. The processing terminal calculates the coordinates of one feature point and subtracts the coordinates of the other feature point from the coordinates of the two feature points corresponding to the connection feature point position, thus obtaining the vector pointing from one point to another.

[0127] Step S402: Determine the standard feature point vector based on the position of the standard feature point.

[0128] The standard feature point vector refers to the vector between the standard disconnector connection parts. The processing terminal calculates the coordinates of one feature point and subtracts the coordinates of the other feature point from the coordinates of the two feature points corresponding to the standard feature point position, thus obtaining the vector pointing from one point to another.

[0129] Step S403: Calculate the vector angle between the connecting feature point vector and the standard feature point vector to generate the torsional deformation of the knife gate.

[0130] In this process, after determining the connecting feature point vector and the standard feature point vector, the processing terminal substitutes the connecting feature point vector and the standard feature point vector into the vector angle formula for calculation, thereby obtaining the torsional deformation degree of the disconnector. The vector angle between the connecting parts before and after the disconnector deformation reflects the degree of torsion of the disconnector, ensuring the accuracy of the torsional deformation degree of the disconnector.

[0131] Reference Figure 5 The steps for generating the telescopic deformation degree of the knife switch based on the set of contour feature points and the preset standard feature point spacing include:

[0132] Step S500: Extract the position of the knife gate end feature point from the contour feature point set.

[0133] Among them, the position of the feature point at the end of the disconnector refers to the coordinates of the feature points at both ends of the disconnector. It is obtained by the processing terminal from the set of contour feature points. By determining the position of the feature point at the end of the disconnector, data support is provided for the subsequent calculation of the length of the current disconnector.

[0134] Step S501: Calculate the distance between the feature points at the end of the disconnector to generate the fixed length of the disconnector.

[0135] The fixed length of the disconnect switch refers to the current length of the disconnect switch, which is calculated by the processing terminal by substituting the position of the feature point at the end of the disconnect switch into the distance formula between the points. By determining the fixed length of the disconnect switch, data support is provided for subsequent comparison with the standard length.

[0136] Step S502: Calculate the difference between the fixed length of the disconnect switch and the standard feature point spacing to generate the disconnect switch telescopic distance.

[0137] The telescopic distance of the disconnector refers to the telescopic length of the current disconnector relative to the standard length, which is obtained by the processing terminal by calculating the absolute value of the difference between the fixed length of the disconnector and the standard feature point spacing.

[0138] Step S503: Calculate the quotient of the disconnector extension distance and the standard feature point spacing to generate the disconnector extension deformation degree.

[0139] In this process, after determining the telescopic distance of the disconnector, the processing terminal calculates the quotient between the telescopic distance of the disconnector and the standard feature point spacing, thereby determining the proportion of the changed length of the disconnector to the standard length, and thus obtaining the telescopic deformation degree of the disconnector.

[0140] Reference Figure 6 The steps for adjusting the preset standard position state mapping relationship according to the degree of deformation of the switch to generate the actual position state mapping relationship include:

[0141] Step S600: Based on the degree of deformation of the disconnector, find the corresponding disconnector deformation influence coefficient in the preset deformation influence relationship.

[0142] Among them, the deformation influence relationship refers to the correspondence between different levels of deformation and the deformation influence coefficient of the disconnector. For example, the coefficient corresponding to the first level of deformation is 0.1, and the coefficient corresponding to the second level of deformation is 0.3, so as to ensure that the greater the degree of deformation, the greater the adjustment influence coefficient. The operator forms a mapping table by matching the different levels of deformation with the deformation influence coefficient of the disconnector.

[0143] The deformation influence coefficient of the disconnect switch refers to the degree of adjustment to the mapping relationship of the standard position state. The larger the deformation influence coefficient of the disconnect switch, the greater the degree of adjustment. It is obtained by the processing terminal by looking up the corresponding mapping table of deformation influence relationship according to the degree of disconnect switch deformation.

[0144] Step S601: Normalize and sum the bending deformation, torsional deformation, and expansion deformation of the disconnector to generate the deformation influence value.

[0145] Among them, the deformation influence value refers to the intensity of the influence of the disconnector deformation on the position of the disconnector feature point. The processing terminal calculates the product of the disconnector bending deformation degree and 100, the disconnector torsional deformation degree and 180 / π, and the disconnector expansion deformation degree and 100. Finally, the three products are added together to obtain the deformation influence value. The deformation influence value maps the deformation parameters of different physical dimensions to a scale, reflecting the degree of influence of the disconnector deformation on the position of the feature point.

[0146] Step S602: Collect and adjust the direction coefficient.

[0147] The adjustment direction coefficient refers to the adjustment direction of the standard feature point position. The processing terminal calculates the distance between the current knife switch profile feature point and the standard profile line. If the distance is greater than 0, it indicates that the feature point is positively deflected towards the normal of the line, and the abscissa of the feature point should be reduced. At this time, the adjustment direction coefficient is -1. If the distance is less than 0, it indicates that the feature point is negatively deflected towards the normal of the line, and the abscissa of the feature point should be increased. At this time, the adjustment direction coefficient is 1.

[0148] Step S603: Calculate the product between the adjustment direction coefficient, the knife gate deformation influence coefficient, and the deformation influence value to generate the position adjustment parameters.

[0149] Among them, the position adjustment parameter refers to the adjustment direction and verticality of the standard feature point position. It is obtained by the processing terminal by calculating the product of the adjustment direction coefficient, the knife gate deformation influence coefficient and the deformation influence value. By calculating the product of the knife gate deformation influence coefficient and the deformation influence value, the specific offset is determined. Then, the adjustment direction coefficient is used to assign the adjustment direction to the offset to ensure the accuracy of the position adjustment parameter.

[0150] Step S604: Adjust the standard position state mapping relationship according to the position adjustment parameters to generate the actual position state mapping relationship.

[0151] In this step, the actual position state mapping relationship is consistent with the actual position state mapping relationship in step S1022. The processing terminal adjusts the position of the feature point in the standard position state mapping relationship according to the adjustment direction and data in the position adjustment parameters, thereby obtaining the actual position state mapping relationship, ensuring that the actual position state mapping relationship is adapted to the deformation of the disconnector.

[0152] Reference Figure 7 The steps for collecting the feature point locations of the disconnector include:

[0153] Step S700: Perform gradient calculation on the pixels in the knife switch detection image according to the preset local window parameters to generate the horizontal gradient and vertical gradient of the feature points.

[0154] The local window parameter refers to a window centered on the pixel coordinates, such as 3×3 or 5×5. By using the local window parameter, the coordinates around the pixel coordinates are isolated, making it easier to calculate the gradient of the pixel coordinates.

[0155] The horizontal gradient of a feature point refers to the gradient of the feature point in the horizontal direction, and the vertical gradient of a feature point refers to the gradient of the feature point in the horizontal direction. They reflect the gray-level change rate of the pixel in the horizontal and vertical directions. The feature point is located at the connection inflection point of the knife gate, where the gray-level change rate is relatively large. By calculating the gradient, data support can be provided for the selection of feature points. The processing terminal separates the pixel image and coordinates in the knife gate detection image according to the local window parameters, and then uses the Sobel operator to calculate the gradient of the pixel image to obtain the horizontal gradient and vertical gradient of the feature point.

[0156] Step S701: Construct the gradient matrix of the feature point based on the horizontal and vertical gradients of the feature point.

[0157] The feature point gradient matrix is ​​a second-order matrix composed of the gradients of feature points. The first row and first column of the matrix is ​​the square of the horizontal gradient of the feature point. The first row and second column and the second row and first column of the matrix are the product of the horizontal gradient and the vertical gradient of the feature point. The second row and second column of the matrix is ​​the square of the vertical gradient of the feature point. By constructing the feature point gradient matrix, the intensity of gray-level changes of the feature point in the horizontal and vertical directions can be reflected by the eigenvalues ​​of the matrix.

[0158] Step S702: Calculate the gradient matrix of the feature points according to the preset response function to generate the feature point response values.

[0159] The response function is a function used to calculate the response value of a feature point. The response function used in Harris corner detection is adopted. Specifically, the matrix determinant is calculated first, then the product of the empirical coefficient and the matrix trace is calculated, and finally the difference between the matrix determinant and the product of the empirical coefficient and the matrix trace is calculated. The intensity of the gray change of the feature point is calculated through the response function, thereby distinguishing whether the feature point is the required feature point.

[0160] The feature point response value is an indicator of whether a feature point is a corner feature. It is obtained by the processing terminal after calculating the feature point gradient matrix based on the response function. If the feature point is a corner point, the gray-level gradient changes in both the horizontal and vertical directions are large, and the value of the matrix determinant is much larger than the product of the matrix trace and the empirical coefficients. Therefore, the feature point response value is a large positive value. If the feature point is an edge point, the gray-level gradient changes in only one direction are large, while the gray-level gradient changes in the other direction are small. Therefore, the value of the matrix determinant is less than the product of the matrix trace and the empirical coefficients. Therefore, the feature point response value is 0 or negative. If the feature point is a flat point, the gray-level gradient changes in both directions are small. Therefore, the matrix determinant and the product of the matrix trace and the empirical coefficients are all close to 0, and the feature point response value is also close to 0.

[0161] Step S703: Filter and sort the feature point response values ​​according to the preset response threshold to determine the maximum response value.

[0162] The maximum response value refers to the maximum value among the feature point response values. The processing terminal first compares the feature point response values ​​with the response threshold to filter out the feature point response values ​​that are greater than the response threshold. These feature points are corner features. Then, the response values ​​of these feature points are sorted, and the largest response value is selected to provide data support for determining the location of the feature points.

[0163] The response threshold is the minimum response value that determines a feature point as a corner feature. The specific value is determined by the operator based on the actual situation.

[0164] Step S704: Collect the initial position of the feature point corresponding to the maximum response value.

[0165] The initial position of the feature point refers to the position of the feature point determined based on its properties. It is determined by the processing terminal when determining the maximum response value by calling the coordinates of the pixel corresponding to the maximum response value. The initial position of the feature point is used to filter the feature points once, which facilitates further filtering to determine a more accurate feature point position.

[0166] Step S705: Match the initial positions of the feature points to generate the feature point positions.

[0167] The feature point positions in this step are consistent with those in step S103. They are obtained by the processing terminal matching the initial positions of the feature points based on the feature points of the standard image. For details, please refer to [link to relevant documentation]. Figure 8 The steps.

[0168] Reference Figure 8 The steps for matching the initial positions of feature points to generate feature point positions include:

[0169] Step S800: Extract the feature point neighborhood image from the knife switch detection image based on the initial position of the feature point and the preset neighborhood image parameters.

[0170] Among them, the feature point neighborhood image refers to the neighborhood image captured with the initial position of the feature point as the center. It is captured by the processing terminal with the initial position of the feature point as the center and the window size corresponding to the neighborhood image parameters in the knife switch detection image. By determining the feature point neighborhood image, it is convenient to compare it with the standard image later.

[0171] The neighborhood image parameter refers to the size of the neighborhood image of the feature point. In this embodiment, 30×30 is used as an example.

[0172] Step S801: Determine the feature point pixels and the average neighborhood pixels based on the feature point neighborhood image.

[0173] Among them, feature point pixels refer to the pixel values ​​of different feature points in the feature point neighborhood image, which are identified by the processing terminal in the feature point neighborhood image. Neighborhood average pixels refer to the average pixel value of all feature points in the feature point neighborhood image, which is calculated by the processing terminal based on the average pixel values ​​of the feature points, providing data support for subsequently determining the similarity between the feature point neighborhood image and the standard image.

[0174] Step S802: Calculate the similarity between feature point pixels, neighborhood average pixels, and preset standard image pixels, and select the maximum value to generate feature point matching degree.

[0175] The standard image pixels refer to the pixels of all feature points in the standard image and the average pixel value, which are obtained by the processing terminal through recognition and calculation of the standard image. This will not be elaborated here. The standard image is the neighborhood image of the standard feature points, and the image size is 15×15.

[0176] Feature point matching degree refers to the degree of matching between the neighborhood image of a feature point and the standard image. The processing terminal uses the standard image as a sliding window to slide and match in the neighborhood image of the feature point. The feature point pixels, the average neighborhood pixels, and the standard image pixels are substituted into the normalized cross-correlation function for calculation to obtain the correlation coefficient between the sliding region image and the standard image. The correlation coefficient ranges from -1 to 1. The closer the correlation coefficient is to 1, the higher the similarity of the images. The largest correlation coefficient is then selected and defined as the feature point matching degree to ensure the accuracy of the feature point position.

[0177] Step S803: Collect the feature point matching position corresponding to the feature point matching degree.

[0178] Among them, the feature point matching position refers to the coordinates of the center point of the sliding region image corresponding to the feature point matching degree. When the processing terminal calculates the correlation coefficient, the coordinates of the center point of all sliding region images are recorded. When determining the maximum correlation coefficient, the coordinates of the center point of the sliding region image corresponding to the maximum correlation coefficient are defined as the feature point matching position.

[0179] Step S804: Determine the position fusion weights based on the feature point matching degree.

[0180] Among them, the position fusion weight refers to the weight of the fused feature point matching position and the initial position of the feature point. The processing terminal calculates the difference between the feature point matching degree and 0.5, and then compares the difference with 0 to select the maximum value between the difference and 0. The value range of the position fusion weight is 0 to 0.5. If the position fusion weight is smaller, it indicates that the correlation coefficient is smaller, and the feature point matching position accounts for a smaller degree.

[0181] Step S805: The matching position and the initial position of the feature point are fused according to the position fusion weight to generate the feature point position.

[0182] In this process, after determining the position fusion weight, the coordinates corresponding to the feature point matching position and the initial position of the feature point are weighted and summed based on the position fusion weight and the difference between 1 and the position fusion weight. This process merges the positions selected in the two steps to obtain the feature point position, thereby improving the accuracy of the feature point position.

[0183] Based on the same inventive concept, embodiments of this application provide a visual inspection system for a knife switch, comprising:

[0184] The acquisition module is used to acquire the knife switch detection image, historical deformation degree, historical position state mapping relationship, feature point position, adjustment direction coefficient, initial position of feature point and feature point matching position;

[0185] Memory, used to store the program of the knife switch-based visual inspection method;

[0186] The processor can load and execute programs in memory to implement a knife switch-based visual inspection method.

[0187] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0188] This application provides a computer-readable storage medium storing a computer program that can be loaded by a processor and executed as a visual inspection method based on a knife switch.

[0189] Computer storage media include, for example, USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media that can store program code.

[0190] Based on the same inventive concept, embodiments of this application provide a smart terminal, including a memory and a processor, wherein the memory stores a computer program that can be loaded and executed by the processor for a visual inspection method based on a knife switch.

[0191] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0192] The above are all preferred embodiments of this application and are not intended to limit the scope of protection of this application. Any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is only one example of a series of equivalent or similar features.

Claims

1. A method of visual inspection based on a blade shutter, characterized in that, The method comprises the following steps: Collecting a blade detection image and a historical deformation degree of a blade; Analyzing the blade detection image to determine the deformation degree of the blade; Determining whether the deformation degree of the blade meets the requirements of the historical deformation degree; If yes, collecting a historical position-state mapping relationship and determining the historical position-state mapping relationship as an actual position-state mapping relationship; If no, adjusting a preset standard position-state mapping relationship according to the deformation degree of the blade to generate an actual position-state mapping relationship; Collecting a feature point position of the blade; Determining a blade detection state according to the feature point position and the actual position-state mapping relationship.

2. The blade-based vision inspection method of claim 1, wherein, The step of analyzing the blade detection image to determine the deformation degree of the blade comprises the following steps: Extracting contour feature points of the blade detection image to generate a contour feature point set; Calculating according to the contour feature point set and a preset standard contour straight line to generate a blade bending deformation degree; Calculating according to the contour feature point set and a preset standard contour point set to generate a blade twisting deformation degree; Calculating according to the contour feature point set and a preset standard feature point spacing to generate a blade stretching deformation degree; Analyzing the blade bending deformation degree, the blade twisting deformation degree and the blade stretching deformation degree according to a preset blade deformation rule library to determine the deformation degree of the blade.

3. The blade-based vision inspection method of claim 2, wherein, The step of calculating according to the contour feature point set and the preset standard contour straight line to generate the blade bending deformation degree comprises the following steps: Traversing a Euclidean distance between the contour feature point set and the standard contour straight line to generate a contour feature point straight line distance; Normalizing the contour feature point straight line distance according to a preset contour feature point straight line maximum distance to generate a contour feature point bending degree; Calculating a mean value of the contour feature point bending degree to generate the blade bending deformation degree.

4. The blade-based vision inspection method of claim 2, wherein, The step of calculating according to the contour feature point set and the preset standard contour point set to generate the blade twisting deformation degree comprises the following steps: Respectively extracting a connection feature point position in the contour feature point set and a standard feature point position in the standard contour point set; Determining a connection feature point vector according to the connection feature point position; Determining a standard feature point vector according to the standard feature point position; Calculating a vector included angle between the connection feature point vector and the standard feature point vector to generate the blade twisting deformation degree.

5. The blade-based vision inspection method of claim 2, wherein, The step of calculating according to the contour feature point set and the preset standard feature point spacing to generate the blade stretching deformation degree comprises the following steps: Extracting a blade end feature point position in the contour feature point set; Calculating a distance between the blade end feature point positions to generate a blade fixed length; Calculating a difference value between the blade fixed length and the standard feature point spacing to generate a blade stretching distance; Calculating a quotient of the blade stretching distance and the standard feature point spacing to generate the blade stretching deformation degree.

6. The blade-based vision inspection method of claim 2, wherein, The step of adjusting the preset standard position-state mapping relationship according to the deformation degree of the blade to generate an actual position-state mapping relationship comprises the following steps: Looking up a corresponding blade deformation influence coefficient in a preset deformation influence relationship according to the deformation degree of the blade; Summing up the blade bending deformation degree, the blade twisting deformation degree and the blade stretching deformation degree to generate a deformation influence value; Collecting an adjustment direction coefficient; The product of the adjustment direction coefficient, the knife gate deformation influence coefficient, and the deformation influence value is calculated to generate a position adjustment parameter; The standard position state mapping relationship is adjusted according to the position adjustment parameter to generate an actual position state mapping relationship.

7. The blade-based vision inspection method of claim 1, wherein, The step of collecting the feature point position of the knife gate includes: Gradient calculation is performed on the pixels in the knife gate detection image according to the preset local window parameter to generate a feature point horizontal gradient and a feature point vertical gradient; A feature point gradient matrix is constructed according to the feature point horizontal gradient and the feature point vertical gradient; The feature point gradient matrix is calculated according to the preset response function to generate a feature point response value; The feature point response value is sorted according to the preset response threshold to determine a maximum response value; The initial position of the feature point corresponding to the maximum response value is collected; The initial position of the feature point is matched to generate a feature point position.

8. The blade-based vision inspection method of claim 7, wherein, The step of matching the initial position of the feature point to generate a feature point position includes: A feature point neighborhood image is extracted from the knife gate detection image according to the initial position of the feature point and the preset neighborhood image parameter; A feature point pixel and a neighborhood average pixel are determined according to the feature point neighborhood image; The similarity between the feature point pixel, the neighborhood average pixel, and the preset standard image pixel is calculated, and the maximum value is filtered to generate a feature point matching degree; The matching position of the feature point corresponding to the feature point matching degree is collected; The position fusion weight is determined according to the feature point matching degree; The matching position of the feature point and the initial position of the feature point are fused according to the position fusion weight to generate the feature point position.

9. A visual inspection system for a blade shutter, the system comprising: a housing; a light source; a camera; a light source controller; a camera controller; and a processor. It includes: The acquisition module is configured to acquire a knife gate detection image, a historical deformation degree, a historical position state mapping relationship, and a feature point position; The memory is configured to store a program of the knife gate-based visual detection method according to any one of claims 1 to 8; The processor, the program in the memory can be loaded and executed by the processor, and the knife gate-based visual detection method according to any one of claims 1 to 8 is implemented.

Images