Machine vision-based intelligent detection method and system for disassembled electric energy meter
Patent Information
- Application Number
- CN202411002680.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2024-11-08
AI Technical Summary
[0005]上述现有技术存在的问题是,仅通过图像识别电表中条形码的信息,对电表信号进行确定,但是在拆回电表中,条形码信息及rfid识别卡信息大多为不准确的,通过此信息进行分类检定存在一定缺陷
[0077] Improve detection efficiency: The present invention uses machine vision technology to realize automatic detection and classification of dismantled electric energy meters, greatly improving detection efficiency.
Smart Images

Figure CN118918299A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of electric meter calibration, and more specifically, to a method and system for intelligently detecting a disassembled electric energy meter based on machine vision. Background Art
[0002] The watt-hour meter calibration device is the main standard instrument used to calibrate the watt-hour meter and realize the transmission and traceability of the watt-hour meter value. With the continuous development of watt-hour meter technology and the extensive application of smart watt-hour meters, the meter calibration quantity has suddenly doubled, and the measurement requirements for watt-hour meters in different occasions have also increased. In actual production applications, a large number of watt-hour meters returned from power supply bureaus, calibration lines, user returns, etc. need to be re-tested to detect deviations after application and determine whether they have "reuse" value.
[0003] The existing testing platform does not have relevant artificial intelligence algorithms, and has poor scalability and compatibility. The relatively cumbersome testing process under the old model also makes it difficult to balance labor costs. Manual testing requires a lot of manpower and material resources, and the test results are easily affected by human factors, and the accuracy and consistency of the test cannot be guaranteed. In addition, manual testing also has problems such as slow detection speed and limited detection range, which cannot meet the needs of large-scale dismantling of electric energy meters for testing.
[0004] Existing technologies such as the Chinese patent with the patent number "CN117826060A" disclose a system and method for calibrating an electric energy meter, a storage medium, and a computer device. The system includes: an intelligent storage module, an automatic loading module, an image processing module, and a performance calibration module; the intelligent storage module is used to respond to the electric energy meter calibration task and obtain the turnover box position of the electric energy meter to be calibrated corresponding to the electric energy meter calibration task; the automatic loading module is used to add the electric energy meter to be calibrated in the turnover box to the transmission tray based on the turnover box position; the image processing module is used to take an image of the electric energy meter to be calibrated in the transmission tray, and extract the barcode information of the electric energy meter to be calibrated based on the captured image; the performance calibration module is used to obtain the type of electric energy meter based on the barcode information, and transmit the electric energy meter to be calibrated to the calibration pipeline for calibrating the electric energy meter type for performance calibration. The electric energy meter is distributed to the corresponding calibration pipeline for automatic calibration through image recognition, which improves the calibration efficiency of the electric energy meter.
[0005] The problem with the above-mentioned prior art is that the meter signal is determined only by image recognition of the barcode information in the meter. However, when the meter is disassembled, the barcode information and RFID identification card information are mostly inaccurate, and there are certain defects in classification and verification based on this information. Summary of the invention
[0006] In order to solve the above technical problems, the present invention proposes a method system for intelligent detection of dismantled electric energy meters based on machine vision.
[0007] The technical solution of the present invention is as follows:
[0008] On the one hand, the present invention provides a machine vision-based intelligent detection system for dismantled electric energy meters, comprising:
[0009] Image acquisition module: The infrared and visible light cameras installed on the assembly line collect images of the disassembled electric meters, and transmit the image information to the image processing module through the communication interface;
[0010] Image processing module: pre-process the collected images, including grayscale, binarization, and denoising operations;
[0011] Feature extraction module: extract key feature points from the processed disassembled meter image using a scale-invariant feature transformation algorithm;
[0012] Classification and recognition module: The extracted key feature points are input into the pre-trained support vector machine model to classify and recognize the returned meters, distinguish single-phase meters from three-phase meters, determine the models and parameters of the returned meters, and send the models and parameter information of the returned meters to the sequential positioning module through the communication interface;
[0013] Sequential positioning module: It realizes accurate navigation and positioning of the returned meters according to the model and parameter information of the returned meters and the magnetic tracks and radio frequency identification cards laid under the assembly line, and generates a queue of returned meter equipment;
[0014] Functional verification module: According to the generated queue of disassembled electric meter equipment, it automatically matches the crimping operation and detection procedures that need to be performed, and outputs the verification results.
[0015] As a preferred implementation, the feature extraction module extracts key feature points from the disassembled electric meter image, including the following steps:
[0016] Scale space extreme value detection: The scale space of the image is simulated by constructing a Gaussian pyramid; at each scale, the local extreme value points are found by comparing the grayscale values of each pixel with its neighboring pixels;
[0017] Key point positioning: accurately locate the detected extreme points; determine the position and scale of the key points by fitting the second-order Taylor expansion;
[0018] Direction assignment: assign a main direction to each key point; calculate the gradient direction histogram of the key point neighborhood and select the peak direction in the histogram as the main direction of the key point;
[0019] Key feature point generation: Generate a descriptor for each key point to describe its local features, and construct the key feature point by calculating the gradient information and direction information of the key point neighborhood.
[0020] As a preferred implementation, the classification and recognition module matches the characteristic values according to the extracted key characteristic points to obtain the model information of the electric meter to be inspected; comprising the following steps:
[0021] Using Euclidean distance as the basis for similarity determination, the extracted key feature points are matched with the model meter image. A key feature point of the model meter image is taken from the two key feature points with the closest Euclidean distance in the disassembled meter image. Among these two key feature points, if the closest distance divided by the second closest Euclidean distance is less than a certain ratio threshold, the match is successful, and the model information corresponding to the model meter image is determined to be the model information of the disassembled meter; the specific formula is as follows:
[0022] Describe the similarity between two key feature points:
[0023]
[0024] In the formula, R i is the key feature point in the model meter image, S i To retrieve the key feature points in the meter image, r ij and ij Respectively represent the key descriptors in the meter image and the key point descriptors in the meter image; j = 1, 2, 3..., 128;
[0025] Then we have:
[0026]
[0027] In the formula, D(R i , S 1 ), D(R i , S 2 ) are respectively the Euclidean distances between a key feature point of the model meter image and the closest and second closest feature key points of the disassembled meter image;
[0028] After obtaining the matching relationship between the key feature points of the model meter image and the disassembled meter image, the model meter image and the disassembled meter image undergo the following transformation process:
[0029]
[0030] Where: H is the projection transformation matrix; [x', y', 1] T and [x,y,1] T are matrices composed of the coordinates of the pixels of a set of feature matching points on the model meter image and the disassembled meter image;
[0031] In order to eliminate the possible erroneous matching points in the matching process, random sampling consistency is used here to exclude erroneous matching points; this is a robust parameter estimation method, which uses randomly selected samples to estimate parameters and find a solution that has the greatest adaptability to the data; under a certain confidence probability P, at least one group of data in the M groups of samples is all internal points, that is:
[0032] 1-(1-ε m ) M =P
[0033] Where: ε is the inlier rate; m is the minimum amount of data required in the model parameter calculation process.
[0034] As a preferred implementation, the pre-training of the support vector machine model comprises the following steps:
[0035] Data preparation: Collect a set of labeled meter image samples to form a data set, including matching pairs and non-matching pairs. Matching pairs are meter images of the same model, and non-matching pairs are meter images of different models. Each sample can be represented as a feature vector.
[0036] Data partitioning: Divide the data set into training set and test set; the training set is used to train the support vector machine model, and the test set is used to evaluate the performance of the model;
[0037] Model training: Use the training set to train the support vector machine model; the support vector machine separates matching pairs from non-matching pairs by finding an optimal hyperplane;
[0038] Model evaluation: Use the test set to evaluate the performance of the trained support vector machine model; various evaluation indicators can be used, such as accuracy and recall;
[0039] Parameter tuning: According to the model evaluation results, adjust the parameters of the support vector machine model, such as the kernel function type and regularization parameter, to improve the performance of the model;
[0040] Image matching: Use the trained support vector machine model to match the images of the dismantled electricity meters to obtain matching results.
[0041] On the other hand, the present invention also provides a method for intelligent detection of dismantled electric energy meters based on machine vision, comprising the following steps:
[0042] Step S1: The image of the disassembled electric meter is collected by an infrared visible light camera installed on the assembly line, and the image information is transmitted to the image processing module through the communication interface;
[0043] Step S2: preprocessing the collected image, including grayscale, binarization, and denoising operations;
[0044] Step S3: extracting key feature points from the processed disassembled electricity meter image using a scale-invariant feature transformation algorithm;
[0045] Step S4: Input the extracted key feature points into a pre-trained support vector machine model to classify and identify the returned meters, distinguish single-phase meters from three-phase meters, determine the models and parameters of the returned meters, and send the models and parameter information of the returned meters to the sequential positioning module through the communication interface;
[0046] Step S5: accurately navigate and locate the returned electric meter according to the model and parameter information of the returned electric meter and the magnetic track and radio frequency identification card laid under the assembly line, and generate a returned electric meter equipment queue;
[0047] Step S6: According to the generated queue of disassembled electric meter equipment, the crimping operation and detection procedure to be performed are automatically matched, and the verification result is output.
[0048] As a preferred implementation, the extraction of key feature points in the disassembled electric meter image comprises the following steps:
[0049] Scale space extreme value detection: The scale space of the image is simulated by constructing a Gaussian pyramid; at each scale, the local extreme value points are found by comparing the grayscale values of each pixel with its neighboring pixels;
[0050] Key point positioning: accurately locate the detected extreme points; determine the position and scale of the key points by fitting the second-order Taylor expansion;
[0051] Direction assignment: assign a main direction to each key point; calculate the gradient direction histogram of the key point neighborhood and select the peak direction in the histogram as the main direction of the key point;
[0052] Key feature point generation: Generate a descriptor for each key point to describe its local features, and construct the key feature point by calculating the gradient information and direction information of the key point neighborhood.
[0053] As a preferred implementation, the model information of the electric meter to be inspected is obtained according to the extracted key feature points and matching feature values; the steps include:
[0054] Using Euclidean distance as the basis for similarity determination, the extracted key feature points are matched with the model meter image. A key feature point of the model meter image is taken from the two key feature points with the closest Euclidean distance in the disassembled meter image. Among these two key feature points, if the closest distance divided by the second closest Euclidean distance is less than a certain ratio threshold, the match is successful, and the model information corresponding to the model meter image is determined to be the model information of the disassembled meter; the specific formula is as follows:
[0055] Describe the similarity between two key feature points:
[0056]
[0057] In the formula, R i is the key feature point in the model meter image, S i To retrieve the key feature points in the meter image, r ij and ij Respectively represent the key descriptors in the meter image and the key point descriptors in the meter image; j = 1, 2, 3..., 128;
[0058] Then we have:
[0059]
[0060] In the formula, D(R i , S 1 ), D(R i , S 2 ) are respectively the Euclidean distances between a key feature point of the model meter image and the closest and second closest feature key points of the disassembled meter image;
[0061] After obtaining the matching relationship between the key feature points of the model meter image and the disassembled meter image, the model meter image and the disassembled meter image undergo the following transformation process:
[0062]
[0063] Where: H is the projection transformation matrix; [x', y', 1] T and [x, y, 1] f are matrices composed of the coordinates of the pixels of a set of feature matching points on the model meter image and the disassembled meter image;
[0064] In order to eliminate the possible erroneous matching points in the matching process, random sampling consistency is used here to exclude erroneous matching points; this is a robust parameter estimation method, which uses randomly selected samples to estimate parameters and find a solution that has the greatest adaptability to the data; under a certain confidence probability P, at least one group of data in the M groups of samples is all internal points, that is:
[0065] 1-(1-ε m ) M =P
[0066] Where: ε is the inlier rate; m is the minimum amount of data required in the model parameter calculation process.
[0067] As a preferred implementation, the pre-training of the support vector machine model comprises the following steps:
[0068] Data preparation: Collect a set of labeled meter image samples to form a data set, including matching pairs and non-matching pairs. Matching pairs are meter images of the same model, and non-matching pairs are meter images of different models. Each sample can be represented as a feature vector.
[0069] Data partitioning: Divide the data set into training set and test set; the training set is used to train the support vector machine model, and the test set is used to evaluate the performance of the model;
[0070] Model training: Use the training set to train the support vector machine model; the support vector machine separates matching pairs from non-matching pairs by finding an optimal hyperplane;
[0071] Model evaluation: Use the test set to evaluate the performance of the trained support vector machine model; various evaluation indicators can be used, such as accuracy and recall;
[0072] Parameter tuning: According to the model evaluation results, adjust the parameters of the support vector machine model, such as the kernel function type and regularization parameter, to improve the performance of the model;
[0073] Image matching: Use the trained support vector machine model to match the images of the dismantled electricity meters to obtain matching results.
[0074] On the other hand, the present invention further provides an electronic device having a computer program stored thereon, wherein when the computer program is executed by a processor, the method for intelligent detection of a dismantled electric energy meter based on machine vision as described in any embodiment of the present invention is implemented.
[0075] On the other hand, the present invention also provides a computer-readable medium for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement a machine vision-based intelligent detection method for dismantled electric energy meters as described in any embodiment of the present invention.
[0076] The present invention has the following beneficial effects:
[0077] Improve detection efficiency: The present invention uses machine vision technology to realize automatic detection and classification of dismantled electric energy meters, greatly improving detection efficiency.
[0078] Improve detection accuracy: The present invention utilizes a scale-invariant feature transformation algorithm and a support vector machine model to extract key feature points in the meter image and perform matching, thereby improving detection accuracy and stability.
[0079] Improve the recycling process: The present invention improves the recycling process of dismantled electric energy meters, realizes the statistical integration of dismantled electric energy meter data, and provides strong support for the management and utilization of dismantled electric energy meters. BRIEF DESCRIPTION OF THE DRAWINGS
[0080] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.
[0081] Figure 1 This is a schematic diagram of the method flow of Example 1;
[0082] Figure 2 is the Gaussian difference scale space;
[0083] Figure 3 is the SIFT feature vector. DETAILED DESCRIPTION
[0084] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0085] It should be understood that the step numbers used in this document are only for convenience of description and are not intended to limit the order in which the steps are executed.
[0086] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.
[0087] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.
[0088] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.
[0089] Embodiment 1:
[0090] In order to make the purpose, technical solutions and advantages of the present invention more clear, the following will be combined with the specific embodiments of the present application and refer to the attached Figure 1 , clearly and completely describe the technical solution of the present invention.
[0091] In order to solve the problems in the prior art, the present invention provides a method for intelligent detection of disassembled electric energy meters based on machine vision, comprising the following steps:
[0092] Step S1: collect images of the disassembled electric meter through an infrared visible light camera installed on the assembly line, and transmit the image information to an image processing module through a communication interface;
[0093] Step S2: preprocessing the collected image, including grayscale, binarization, denoising and other operations, to improve the image quality and the accuracy of feature extraction;
[0094] Step S3: extracting key feature points from the processed disassembled meter image using a scale-invariant feature transformation algorithm; these key feature points are scale-invariant and rotation-invariant, and can effectively describe the features of meters of different models;
[0095] Extracting key feature points from the disassembled meter image includes the following steps:
[0096] Scale space extreme value detection: The scale space of the image is simulated by constructing a Gaussian pyramid; at each scale, the local extreme value points are found by comparing the grayscale values of each pixel with its neighboring pixels;
[0097] The generation formula of scale space is as follows:
[0098]
[0099] The formula for establishing the DOG Gaussian difference pyramid is as follows:
[0100] D(x,y,mσ)=(G(x,y,k m σ)-G(x,y,k m-1 σ)I(x,y))
[0101] Where: σ is the scale space factor, the larger the σ value, the smoother the image; (x, y) is the position coordinate of a pixel in the image; I is the image formed by the reflected light captured by the image acquisition device; m = 1, 2, ..., s;
[0102]
[0103] Local extrema detection:
[0104] By detecting the extreme values in the scale space, the location and scale of the key points are preliminarily determined. Figure 2 are the three adjacent scales in the Gaussian difference scale space. When detecting the extreme values in the scale space, the pixels marked with crosses in the figure need to be compared with the 26 pixels of the same and adjacent scales to ensure that the extreme points are detected in both the scale space and the two-dimensional image space.
[0105] Key point positioning: Accurately locate the detected extreme points; determine the position and scale of the key points by fitting the second-order Taylor expansion; remove low-contrast key points and unstable edge response points to enhance matching stability and improve noise resistance.
[0106] Direction assignment: assign a main direction to each key point; calculate the gradient direction histogram of the key point neighborhood and select the peak direction in the histogram as the main direction of the key point;
[0107] The gradient direction distribution characteristics of the pixels in the neighborhood of the key point are used to specify the direction parameters for each key point, so that the operator has rotation invariance.
[0108] The calculation formula of the gradient amplitude m(x,y) is:
[0109]
[0110] The calculation formula for the gradient direction θ(x,y) is:
[0111]
[0112] Where L(x,y) represents the pixel value of the image at coordinate (x,y).
[0113] Key feature point generation: Generate a descriptor for each key point to describe its local features, and construct the key feature point by calculating the gradient information and direction information of the key point neighborhood.
[0114] The 8×8 window centered on the key point is divided into 4 4×4 blocks, and the gradient direction histograms in 8 directions are calculated on each 4×4 block. The accumulated value of each gradient direction is plotted to form a seed point, such as Figure 3 In the actual calculation process shown in the figure, in order to enhance the robustness of the matching, only 16 seed points of 4×4 are used to describe each key point, so that a 128-dimensional SIFT feature vector is finally formed for a key feature point. At this time, the SIFT feature vector has removed the influence of geometric deformation factors such as scale change and rotation. If the length of the feature vector is normalized, the influence of illumination change can be further removed.
[0115] Step S4: Input the extracted key feature points into a pre-trained support vector machine model to classify and identify the returned meters, distinguish single-phase meters from three-phase meters, determine the models and parameters of the returned meters, and send the models and parameter information of the returned meters to the sequential positioning module through the communication interface;
[0116] The pre-training of the support vector machine model consists of the following steps:
[0117] Data preparation: Collect a set of labeled meter image samples to form a data set, including matching pairs and non-matching pairs. Matching pairs are meter images of the same model, and non-matching pairs are meter images of different models. Each sample can be represented as a feature vector.
[0118] Data partitioning: Divide the data set into training set and test set; the training set is used to train the support vector machine model, and the test set is used to evaluate the performance of the model;
[0119] Model training: Use the training set to train the support vector machine model; the support vector machine separates matching pairs from non-matching pairs by finding an optimal hyperplane; during the training process, the support vector machine finds the parameters of this optimal hyperplane through an optimization algorithm;
[0120] The optimization algorithm formula is:
[0121]
[0122] The constraints are:
[0123] y i (ω T x i +b)≥1
[0124] Where ω is the normal vector of the hyperplane, b is the intercept, and x i is the feature vector of the i-th sample, y i is the corresponding category label, i = 1, 2, 3…n, and n is the number of samples.
[0125] Model evaluation: Use the test set to evaluate the performance of the trained support vector machine model; various evaluation indicators can be used, such as accuracy and recall;
[0126] The accuracy formula is:
[0127]
[0128] The recall formula is:
[0129]
[0130] TP (True Positive) represents true positive examples, that is, the number of matching pairs correctly predicted as matching pairs; TN (True Negative) represents true negative examples, that is, the number of non-matching pairs correctly predicted as non-matching pairs; FP (False Positive) represents false positive examples, that is, the number of non-matching pairs incorrectly predicted as matching pairs; FN (False Negative) represents false negative examples, that is, the number of matching pairs incorrectly predicted as non-matching pairs.
[0131] Parameter tuning: According to the model evaluation results, adjust the parameters of the support vector machine model, such as the kernel function type and regularization parameter, to improve the performance of the model;
[0132] Image matching: Use the trained support vector machine model to match the images of the dismantled electricity meters to obtain matching results.
[0133] Key feature point matching includes the following steps:
[0134] Using Euclidean distance as the basis for similarity determination, the extracted key feature points are matched with the model meter image. A key feature point of the model meter image is taken from the two key feature points with the closest Euclidean distance in the disassembled meter image. Among these two key feature points, if the closest distance divided by the second closest Euclidean distance is less than a certain ratio threshold, the match is successful, and the model information corresponding to the model meter image is determined to be the model information of the disassembled meter; the specific formula is as follows:
[0135] Describe the similarity between two key feature points:
[0136]
[0137] In the formula, R i is the key feature point in the model meter image, S i To retrieve the key feature points in the meter image, r ij and ij Respectively represent the key descriptors in the meter image and the key point descriptors in the meter image; j = 1, 2, 3..., 128;
[0138] Then we have:
[0139]
[0140] In the formula, D(R i , S 1 ), D(R i , S 2 ) are respectively the Euclidean distances between a key feature point of the model meter image and the closest and second closest feature key points of the disassembled meter image;
[0141] After obtaining the matching relationship between the key feature points of the model meter image and the disassembled meter image, the model meter image and the disassembled meter image undergo the following transformation process:
[0142]
[0143] Where: H is the projection transformation matrix; [x', y', 1] T and [x, y, 1] Tare matrices composed of the coordinates of the pixels of a set of feature matching points on the model meter image and the disassembled meter image;
[0144] In order to eliminate the possible erroneous matching points in the matching process, random sampling consistency is used here to exclude erroneous matching points; this is a robust parameter estimation method, which uses randomly selected samples to estimate parameters and find a solution that has the greatest adaptability to the data; under a certain confidence probability P, at least one group of data in the M groups of samples is all internal points, that is:
[0145] 1-(1-ε m ) M =P
[0146] Where: ε is the internal point rate; m is the minimum amount of data required in the process of model parameter calculation; based on engineering experience, m is taken as 4.
[0147] The appearance features of single-phase and three-phase meters are significantly different. The SIFT algorithm is used to detect key feature points in meter images, such as different meter cover structures, display screen positions, etc. The detected key points are described to generate feature vectors or descriptors, which can express the local appearance and structural information of each key feature point. The key feature points in the two meter images are matched, and the same key feature points or areas are determined by comparing their descriptors to realize the recognition of single-phase and three-phase meters.
[0148] Step S5: accurately navigate and locate the returned electric meter according to the model and parameter information of the returned electric meter and the magnetic track and radio frequency identification card laid under the assembly line, and generate a returned electric meter equipment queue;
[0149] Step S6: According to the generated queue of disassembled electric meter equipment, the crimping operation and detection procedure to be performed are automatically matched, and the verification result is output.
[0150] The electric energy meter data is statistically integrated according to steps S5 and S6, providing strong support for the management and utilization of the removed electric energy meters.
[0151] Embodiment 2:
[0152] This embodiment provides a machine vision-based intelligent detection system for dismantled electric energy meters, including:
[0153] Image acquisition module: collects images of the dismantled electric meters through the infrared and visible light camera installed on the assembly line, and transmits the image information to the image processing module through the communication interface; this module is used to implement the function of step S1 in Example 1, which will not be repeated here.
[0154] Image processing module: pre-processes the collected image, including grayscale, binarization, and denoising operations; this module is used to implement the function of step S200 in Example 1, which will not be repeated here.
[0155] Feature extraction module: extract key feature points from the processed reassembled meter image using a scale-invariant feature transformation algorithm; this module is used to implement the function of step S3 in Example 1, which will not be described in detail here.
[0156] Classification and identification module: input the extracted key feature points into a pre-trained support vector machine model to classify and identify the returned meters, distinguish single-phase meters from three-phase meters, determine the models and parameters of the returned meters, and send the model and parameter information of the returned meters to the sequential positioning module through the communication interface; this module is used to implement the function of step S4 in Example 1, which will not be repeated here.
[0157] Sequential positioning module: accurate navigation and positioning of the returned meters are achieved based on the model and parameter information of the returned meters and the magnetic tracks and radio frequency identification cards laid under the assembly line, and a queue of returned meter equipment is generated; this module is used to implement the function of step S5 in Example 1, which will not be repeated here.
[0158] Function verification module: according to the generated disassembled electric meter equipment queue, automatically matches the crimping operation and detection procedure to be executed, and outputs the verification result. This module is used to implement the function of step S6 in embodiment 1, which will not be described in detail here.
[0159] Embodiment three:
[0160] This embodiment provides an electronic device having a computer program stored thereon. When the computer program is executed by a processor, a method for intelligently detecting a dismantled electric energy meter based on machine vision as described in any embodiment of the present invention is implemented.
[0161] Embodiment 4:
[0162] This embodiment provides a computer-readable medium for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement a machine vision-based intelligent detection method for disassembled electric energy meters as described in any embodiment of the present invention.
[0163] In the embodiments of the present application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Among them, A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c can be single or multiple.
[0164] Those of ordinary skill in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented in a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.
[0165] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0166] In several embodiments provided in the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory; hereinafter referred to as: ROM), random access memory (Random Access Memory; hereinafter referred to as: RAM), disk or optical disk, and other media that can store program codes.
[0167] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. A machine vision-based intelligent detection system for dismantled electric energy meters, characterized in that: include: Image acquisition module: The infrared and visible light cameras installed on the assembly line collect images of the disassembled electric meters, and transmit the image information to the image processing module through the communication interface; Image processing module: pre-process the collected images, including grayscale, binarization, and denoising operations; Feature extraction module: extract key feature points from the processed disassembled meter image using a scale-invariant feature transformation algorithm; Classification and recognition module: The extracted key feature points are input into the pre-trained support vector machine model to classify and recognize the returned meters, distinguish single-phase meters from three-phase meters, determine the models and parameters of the returned meters, and send the models and parameter information of the returned meters to the sequential positioning module through the communication interface; Sequential positioning module: It realizes accurate navigation and positioning of the returned meters according to the model and parameter information of the returned meters and the magnetic tracks and radio frequency identification cards laid under the assembly line, and generates a queue of returned meter equipment; Functional verification module: According to the generated queue of disassembled electric meter equipment, it automatically matches the crimping operation and detection procedures that need to be performed, and outputs the verification results.
2. According to claim 1, a machine vision-based intelligent detection system for dismantled electric energy meters is characterized by: The feature extraction module extracts key feature points in the disassembled electric meter image, including the following steps: Scale space extreme value detection: The scale space of the image is simulated by constructing a Gaussian pyramid; at each scale, the local extreme value points are found by comparing the grayscale values of each pixel with its neighboring pixels; Key point positioning: Accurately locate the detected local extreme points; determine the position and scale of the key points by fitting the second-order Taylor expansion; Direction assignment: assign a main direction to each key point; calculate the gradient direction histogram of the key point neighborhood and select the peak direction in the histogram as the main direction of the key point; Key feature point generation: Generate a descriptor for each key point to describe its local features; construct the key feature point by calculating the gradient information and direction information of the key point neighborhood.
3. According to claim 1, a machine vision-based intelligent detection system for dismantled electric energy meters is characterized by: The classification and recognition module matches the characteristic values according to the extracted key characteristic points to obtain the model information of the electric meter to be inspected; the steps include: Using Euclidean distance as the basis for similarity determination, the extracted key feature points are matched with the model meter image. A key feature point of the model meter image is taken from the two key feature points with the closest Euclidean distance in the disassembled meter image. Among these two key feature points, if the closest distance divided by the second closest Euclidean distance is less than a certain ratio threshold, the match is successful, and the model information corresponding to the model meter image is determined to be the model information of the disassembled meter; the specific formula is as follows: Describe the similarity between two key feature points: In the formula, R i is the key feature point in the model meter image, S i To retrieve the key feature points in the meter image, r ij and ij Respectively represent the key descriptors in the meter image and the key point descriptors in the disassembled meter image; j = 1, 2, 3…, 128; Then we have: In the formula, D(R i ,S1),D(R i ,S2) are the Euclidean distances between a key feature point of the model meter image and the closest and second closest feature key points of the disassembled meter image; After obtaining the matching relationship between the key feature points of the model meter image and the disassembled meter image, the model meter image and the disassembled meter image undergo the following transformation process: Where: H is the projection transformation matrix; [x,,y,,1] T and [x,y,1] T are matrices composed of the coordinates of the pixels of a set of feature matching points on the model meter image and the disassembled meter image; In order to eliminate the possible erroneous matching points in the matching process, random sampling consistency is used to exclude erroneous matching points; the parameters are estimated using randomly selected samples to find a solution that has the greatest adaptability to the data; under a certain confidence probability P, at least one group of data in the M groups of samples is all internal points, that is: 1-(1-e m ) M =P Where: ε is the inlier rate; m is the minimum amount of data required in the model parameter calculation process.
4. According to claim 1, a machine vision-based intelligent detection system for dismantled electric energy meters is characterized by: The pre-training of the support vector machine model includes the following steps: Data preparation: Collect a set of labeled meter image samples to form a data set, which includes matching pairs and non-matching pairs. Matching pairs are meter images of the same model, and non-matching pairs are meter images of different models. Each meter image sample is represented as a feature vector. Data partitioning: Divide the data set into training set and test set; the training set is used to train the support vector machine model, and the test set is used to evaluate the performance of the model; Model training: Use the training set to train the support vector machine model; the support vector machine separates matching pairs from non-matching pairs by finding an optimal hyperplane; Model evaluation: Use the test set to evaluate the performance of the trained support vector machine model; use various evaluation indicators such as accuracy and recall; Parameter tuning: According to the model evaluation results, adjust the parameters of the support vector machine model, such as the kernel function type and regularization parameter, to improve the performance of the model; Image matching: Use the trained support vector machine model to match the images of the dismantled electricity meters to obtain matching results.
5. A method for intelligent detection of dismantled electric energy meters based on machine vision, characterized in that: The following steps are involved: Step S1: The image of the disassembled electric meter is collected by an infrared visible light camera installed on the assembly line, and the image information is transmitted to the image processing module through the communication interface; Step S2: preprocessing the collected image, including grayscale, binarization, and denoising operations; Step S3: extracting key feature points from the processed disassembled electricity meter image using a scale-invariant feature transformation algorithm; Step S4: Input the extracted key feature points into a pre-trained support vector machine model to classify and identify the returned meters, distinguish single-phase meters from three-phase meters, determine the models and parameters of the returned meters, and send the models and parameter information of the returned meters to the sequential positioning module through the communication interface; Step S5: accurately navigate and locate the returned electric meter according to the model and parameter information of the returned electric meter and the magnetic track and radio frequency identification card laid under the assembly line, and generate a returned electric meter equipment queue; Step S6: According to the generated queue of disassembled electric meter equipment, the crimping operation and detection procedure to be performed are automatically matched, and the verification result is output.
6. The method for intelligent detection of dismantled electric energy meters based on machine vision according to claim 5 is characterized in that: The method of extracting key feature points from the disassembled electric meter image comprises the following steps: Scale space extreme value detection: The scale space of the image is simulated by constructing a Gaussian pyramid; at each scale, the local extreme value points are found by comparing the grayscale values of each pixel with its neighboring pixels; Key point positioning: accurately locate the detected extreme points; determine the position and scale of the key points by fitting the second-order Taylor expansion; Direction assignment: assign a main direction to each key point; calculate the gradient direction histogram of the key point neighborhood and select the peak direction in the histogram as the main direction of the key point; Key feature point generation: Generate a descriptor for each key point to describe its local features, and construct the key feature point by calculating the gradient information and direction information of the key point neighborhood.
7. The method for intelligent detection of dismantled electric energy meters based on machine vision according to claim 5 is characterized in that: The method of matching the characteristic values according to the extracted key characteristic points to obtain the model information of the electric meter to be inspected comprises the following steps: Using Euclidean distance as the basis for similarity determination, the extracted key feature points are matched with the model meter image. A key feature point of the model meter image is taken from the two key feature points with the closest Euclidean distance in the disassembled meter image. Among these two key feature points, if the closest distance divided by the second closest Euclidean distance is less than a certain ratio threshold, the match is successful, and the model information corresponding to the model meter image is determined to be the model information of the disassembled meter; the specific formula is as follows: Describe the similarity between two key feature points: In the formula, R i is the key feature point in the model meter image, S i To retrieve the key feature points in the meter image, r ij and ij Respectively represent the key descriptors in the meter image and the key point descriptors in the disassembled meter image; j = 1, 2, 3…, 128; Then we have: In the formula, D(R i ,S1),D(R i ,S2) are the Euclidean distances between a key feature point of the model meter image and the closest and second closest feature key points of the disassembled meter image; After obtaining the matching relationship between the key feature points of the model meter image and the disassembled meter image, the model meter image and the disassembled meter image undergo the following transformation process: Where: H is the projection transformation matrix; [x,,y,,1] T and [x,y,1] T are matrices composed of the coordinates of the pixels of a set of feature matching points on the model meter image and the disassembled meter image; In order to eliminate the possible erroneous matching points in the matching process, random sampling consistency is used here to exclude erroneous matching points; this is a robust parameter estimation method, which uses randomly selected samples to estimate parameters and find a solution that has the greatest adaptability to the data; under a certain confidence probability P, at least one group of data in the M groups of samples is all internal points, that is: 1-(1-e m ) M =P Where: ε is the inlier rate; m is the minimum amount of data required in the model parameter calculation process.
8. The method for intelligent detection of dismantled electric energy meters based on machine vision according to claim 5 is characterized in that: The pre-training of the support vector machine model includes the following steps: Data preparation: Collect a set of labeled meter image samples to form a data set, including matching pairs and non-matching pairs. Matching pairs are meter images of the same model, and non-matching pairs are meter images of different models. Each sample can be represented as a feature vector. Data partitioning: Divide the data set into training set and test set; the training set is used to train the support vector machine model, and the test set is used to evaluate the performance of the model; Model training: Use the training set to train the support vector machine model; the support vector machine separates matching pairs from non-matching pairs by finding an optimal hyperplane; Model evaluation: Use the test set to evaluate the performance of the trained support vector machine model; various evaluation indicators can be used, such as accuracy and recall; Parameter tuning: According to the model evaluation results, adjust the parameters of the support vector machine model, such as the kernel function type and regularization parameter, to improve the performance of the model; Image matching: Use the trained support vector machine model to match the images of the dismantled electricity meters to obtain matching results.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method for intelligent detection of dismantled electric energy meters based on machine vision as described in any one of claims 5 to 8 is implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for intelligent detection of disassembled electric energy meters based on machine vision as described in any one of claims 5 to 8 is implemented.
Citation Information
Patent Citations
AGV control system and control method, and AGV system
CN107450549A
Electric energy representation number and label automatic identification method combining deep neural network and SIFT features
CN110659637A
PSO-SVM-based precision machining structural part identification and classification method
CN112418317A
Automatic sorting system for disassembled electric energy meters
CN114308719A
Ancient book mathematical symbol identification method based on pre-classification
CN116343232A
