Universal pointer meter reading method and device

Through the combination of rotary coordinate box detection and binary graph algorithm, the problem of OCR identification missed detection and false detection in deep learning algorithm is solved, and the accuracy and robustness of meter readings are achieved. It is suitable for a variety of scenarios and equipment conditions. Reduced sample size requirement.

CN118968033APending Publication Date: 2024-11-15WEICHENG INTELLIGENT POWER TECH (HANGZHOU) CO LTD

Patent Information

Application Number
CN202411151861.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-21
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

In the prior art, the meter reading recognition based on deep learning algorithms has problems such as missed and misdetection of OCR recognition, resulting in inaccurate readings and large sample sizes. Due to the environment and equipment conditions, it is difficult to effectively apply in all scenarios.

Method used

Rotating coordinate box detection and OCR character recognition are used, combined with binary graph algorithm to match the scale and scale values, use the preset database to complete the scale values, and calculate the angle information through the pointer tip and scale, calculate the table reading, perform fault-tolerant processing and maximum Long common subsequence matching to improve identification accuracy.

Benefits of technology

It improves the accuracy and robustness of the meter readings, is suitable for different scenarios and equipment conditions, reduces the sample size requirement, enhances the generalization and robustness of the algorithm, and reduces the dependence of image correction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118968033A_ABST
    Figure CN118968033A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of reading identification of meters, solves the problems of inaccurate meter reading caused by omission and false detection of OCR (Optical Character Recognition) identification and large sampling sample size caused by improvement of meter reading accuracy in the prior art, and discloses a reading method and device of a universal pointer meter. Comprising the following steps: carrying out scale and pointer rotation coordinate frame detection and OCR character recognition on a dial plate image, and carrying out one-to-one matching on scales and scale values by using a bipartite graph algorithm; completing the completion of the scale values based on a preset scale value database; the average value of the coordinates of the intersection points of the pointer and the extension lines of the scales is used as the coordinate of the rotation center of the pointer, the angle information of the scales and the pointer is calculated, the reading of the meter is calculated according to the included angle between the pointer and the adjacent scales on the two sides, and high robustness can be achieved for shooting angles, image blurring, illumination, meter smudginess and aged meters; and the success rate and the accuracy of reading identification can be greatly improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of meter reading recognition, and in particular to a method and device for reading a universal pointer meter. Background Art

[0002] At present, meter reading algorithms cover two categories: traditional image algorithms and deep learning algorithms. Traditional algorithms usually use the OpenCV library to perform a series of image processing, such as sharpening, morphological processing, and binarization, and finally derive the readings through rule deduction. Traditional image optimization often requires multiple parameter debugging, which is a large workload, and the parameter application range is relatively limited. In contrast, deep learning algorithms based on dial scale and pointer instance segmentation can accurately locate the dial and pointer, and then analyze the scale information to finally achieve meter readings. Compared with traditional image algorithms, deep learning has a wider applicability and a more convenient optimization iteration process.

[0003] In practical applications, although the reading recognition based on deep learning algorithms is more flexible, it is limited by the camera placement and the environment in which the equipment is located. There are a lot of blurring problems caused by non-straight shooting and long shooting distance, as well as ambient light interference, dirt and aging of the dial itself. When recognizing text on the dial, there are often problems of missed detection and false detection. The inaccurate OCR recognition seriously affects the accuracy of the general meter reading algorithm based on OCR results. Although the problem of false detection and missed detection can be alleviated to a certain extent by collecting poor-quality meter images and optimizing the OCR recognition model, a large number of sample meter images need to be collected. In addition, in some scenarios, difficult samples cannot be collected in large quantities due to objective conditions. This seriously restricts the application of the general meter recognition algorithm based on deep learning algorithms in this scenario. Summary of the invention

[0004] The purpose of the present application is to overcome the problems in the prior art of inaccurate meter readings caused by missed or false detections in OCR recognition and the large amount of sampling samples required to improve the accuracy of meter readings, and to provide a method and device for reading a universal pointer meter.

[0005] In a first aspect, a method for reading a universal pointer meter is provided, comprising:

[0006] Acquire a dial image, wherein the dial image includes scales and pointers;

[0007] Performing rotation coordinate frame detection and OCR character recognition on the dial image for the scale and pointer to obtain the coordinates of the four vertices of the rotation coordinate frame, the position information of the pointer and the scale, and the scale value and the scale value position;

[0008] Use bipartite graph algorithm to perform one-to-one matching between scales and scale values;

[0009] Complete the scale value based on the preset scale value database;

[0010] The average value of the coordinates of the intersection of the pointer and the extended line of the scale is taken as the coordinates of the pointer rotation center;

[0011] Calculate the angle information of the scale and the pointer according to the center point coordinates of the two vertices farthest from the pointer selection center point in the coordinate frame and the pointer rotation center coordinates;

[0012] The meter reading is calculated based on the angle between the pointer and the adjacent scales on both sides.

[0013] Furthermore, the acquisition of the dial image includes:

[0014] The deep learning object detection model is trained through the annotated original dial image to obtain a dial detection model;

[0015] The dial detection model is used to detect the image to be identified to accurately locate the position of the dial in the image to be identified, and then the dial image is extracted.

[0016] Furthermore, after the OCR character recognition, the scale value of the OCR character recognition is subjected to fault tolerance processing, specifically including:

[0017] Constructing a letter-number correspondence database, wherein the letter-number correspondence database contains letter and number combinations that are easily confused in OCR character recognition;

[0018] It is determined whether the scale value is a letter. If the determination result is yes, the letter is corrected to a corresponding number in a letter-number correspondence database.

[0019] Furthermore, a bipartite graph algorithm is used to perform a one-to-one matching between scales and scale values, including: associating nearby scale values ​​for each scale, and completing the one-to-one matching between scales and scale values ​​by minimizing the total distance after matching, wherein the total distance is the sum of the Euclidean distances between all scales and corresponding scale values.

[0020] Furthermore, the scale value is completed based on the preset scale value database, including:

[0021] Building a scale value database according to the business scenario of the meter, wherein the scale value database includes a known scale value sequence of the meter in the business scenario;

[0022] According to the longest common subsequence length of the first sequence of scale values ​​and the second sequence in the scale value database, a complete scale value sequence is matched from the scale value database to complete the scale value completion.

[0023] Furthermore, the first sequence is compared with the second sequence according to the longest common subsequence length of the first sequence composed of scale values ​​and the second sequence in the scale value database. If the first sequence contains a scale value that does not appear in the second sequence, the scale value is replaced with a placeholder, wherein the placeholder can match any number.

[0024] Furthermore, the calculation formula of the meter reading is:

[0025]

[0026] Among them, V p The angle of scale K1 is A1, and the scale value is V1; the angle of scale K2 is A2, and the scale value is V2. The angle of the pointer between scales K1 and K2 is A p .

[0027] In a second aspect, a computer program product comprising instructions is provided, which, when executed on a computer, enables the computer to execute the method described in the first aspect.

[0028] In a third aspect, a computer-readable storage medium is provided, wherein the computer-readable medium stores a program code for execution by a device, wherein the program code includes steps for executing the method in any one of the implementations of the first aspect described above.

[0029] In a fourth aspect, an electronic device is provided, comprising a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements a method as in any one of the implementations in the first aspect above.

[0030] The present application has the following beneficial effects: the present application utilizes the detected rotating coordinate frame to quickly calculate the rotation center of the pointer, thereby avoiding image correction and improving the generalization of the present application; utilizes a bipartite graph algorithm to match the OCR result with a known sequence of dial scale values ​​to obtain each scale value, thereby effectively avoiding the problem of reading calculation errors caused by missed OCR characters and improving the robustness of the algorithm; utilizes the pointer tip, scale and the rotation center of the pointer to calculate the precise angle of the pointer and the scale, thereby effectively improving the reading accuracy, thereby being able to have strong robustness against meter images with missing or blurred scale values ​​caused by shooting angle, image blur, lighting, meter dirt and aging, etc., and is suitable for various meter recognition algorithms, and can greatly improve the success rate and accuracy of reading recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application.

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0033] Figure 1 is a flow chart of a method for reading a universal pointer meter according to Example 1 of the present application;

[0034] Figure 2 This is a comparison diagram of conventional target detection and rotation coordinate frame detection;

[0035] Figure 3 It is a schematic diagram of the positioning of the rotation center of the pointer in the reading method of the universal pointer meter of Example 1 of the present application;

[0036] Figure 4 It is a flow chart of scale value completion in the reading method of the universal pointer meter in Example 1 of the present application;

[0037] Figure 5 It is a schematic diagram of the internal structure of the electronic device of Example 4 of the present application. DETAILED DESCRIPTION

[0038] 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.

[0039] Example 1

[0040] A method for reading a universal pointer meter involved in Example 1 of the present application includes: obtaining a dial image, wherein the dial image contains scales and a pointer; performing rotation coordinate frame detection and OCR character recognition on the dial image to obtain the coordinates of the four vertices of the rotation coordinate frame, the position information of the pointer and the scale, and the scale value and the scale value position; using a bipartite graph algorithm to perform a one-to-one match between the scale and the scale value; completing the scale value based on a preset scale value database; taking the average value of the coordinates of the intersection of the pointer and the extension line of the scale as the coordinates of the pointer rotation center; selecting the center according to the distance from the pointer in the coordinate frame. The angle information of the scale and the pointer is calculated by using the center point coordinates of the two farthest vertices and the coordinates of the pointer rotation center; the meter reading is calculated based on the angle between the pointer and the adjacent scales on both sides. This method uses the detected rotation coordinate frame to quickly calculate the rotation center of the pointer, avoiding image correction and improving the generalization of this application; the OCR result is matched with the known dial scale value sequence by using the bipartite graph algorithm to obtain each scale value, effectively avoiding the reading calculation error caused by OCR character omission and improving the robustness of the algorithm; the precise angle of the pointer and the scale is calculated by using the pointer tip, the scale and the rotation center of the pointer, effectively improving the reading accuracy.

[0041] Specifically, Figure 1 A flow chart showing a method for reading a universal pointer meter in application embodiment 1 is shown, comprising:

[0042] In terms of meter detection, we use a common one-stage deep learning target detection algorithm, such as YoloV5. We train a dial detection model by annotating a large number of dial images. This dial detection model can effectively detect the dial and obtain the dial image by cutting out the parts outside the dial, and it has good generalization. Through dial detection, we can effectively eliminate the interference of the texture outside the dial on the dial reading, and improve the algorithm recognition rate.

[0043] In terms of pointer and scale detection, conventional target detection is generally marked with a horizontal and vertical rectangular frame. This method loses the angle information of the pointer and scale. Therefore, in this embodiment, a rectangular frame with rotation angle information is used to detect the pointer and scale (i.e., rotation coordinate frame detection). This method can not only accurately obtain the position information of the pointer and scale, but also obtain the coordinates of the four vertices of the rotation coordinate frame, and then obtain the high-precision angle information of the scale and pointer, such as Figure 2 shown.

[0044] The extension line of the scale on the dial usually passes through the center of the pointer. If the angle information of the scale and the pointer is known, the extension line can be drawn along the scale direction, such as Figure 3As shown in the figure, the intersection of the extension line and the pointer is the center of the pointer. To reduce the center position error, select the scale with a larger angle with the pointer to extend it, and take the average value of the coordinates of their intersection points as the coordinates of the pointer rotation center. For example, the coordinates of the intersection points are (1, 0), (1.2, 0) and (1, 0.3), then their average value is That is, the coordinates of the pointer's rotation center are (1.1, 0.1).

[0045] Use commonly used OCR character recognition algorithms, such as paddleocr, to identify various characters on the dial. The results include the value and position information of the characters, and build a letter-number correspondence database, where the letter-number correspondence database contains letter and number combinations that are easily confused in OCR character recognition, such as O-0.I-1, l-1, etc., and then perform fault tolerance processing on easily confused characters, for example: the letter 'O' is modified to the number 0, the letters 'I' and 'l' are modified to the number 1, etc.

[0046] The method of using the bipartite graph algorithm to perform one-to-one matching between scales (i.e. scale lines) and scale values ​​and to complete the scale values ​​is as follows: Figure 4 shown.

[0047] Matching scales and scale values: The quality of the match is measured by the sum of the distances between the matched scales and scale values. This is equivalent to the bipartite graph problem. The sum of the matching edges is minimized through the Hungarian matching algorithm. The value of the matching edge is the Euclidean distance between the scale and the scale value.

[0048] Specifically, the scale and scale value are matched: each scale is associated with a nearby scale value, and the scale and scale value are matched by minimizing the total distance after matching (the sum of the distances of the global scale and scale value matching).

[0049] For example, 5 scales (a, b, c, d, e) and 4 scale values ​​(0, 1, 2, 4) are identified. Because of the problem of inaccurate identification, they need to be matched one-to-one. According to the Euclidean distance of the coordinates of the 5 scales and the 4 scale values ​​on the graph, a matching relationship is generated, and the final result may be (a, 2)(b, 4)(c, 2)(e, 0). The global distance is minimized, that is, this matching combination is the combination with the smallest matching distance among all matches.

[0050] Completing scale values: First, assuming that we already know that the sequence of different dial scales is a finite set, at least in a fixed business scenario, it must be exhaustive. Next, we use the set of these scale sequences (i.e., the scale value database) to assist in completing the detected missing scale value sequence.

[0051] The indicator to measure the quality of completion is the maximum a posteriori probability of the current missing scale value appearing based on a certain candidate scale value sequence. The larger the posterior probability, the greater the possibility that the candidate scale value sequence is the target sequence to be completed. The calculation formula of the maximum a posteriori probability is:

[0052]

[0053] Where L is the observation sequence, li is the legal target sequence in the database, P(L) is the occurrence probability of the observation sequence, To maximize the probability of the observed sequence, P(l i ) is the marginal probability of a target sequence in the database, P(L|l i ) is the likelihood estimate of observing L when the dial scale sequence is li, log(length(l i )) is the value of P(l i ), because the longer the sequence, the easier it is to match lcs of the same length, but the probability of such li appearing will be smaller.

[0054] It should be noted that there are n,...n target sequences in the scale value database, and we need to select a target sequence that is most likely to be identified as the observation sequence. At the same time, the overall probability of occurrence of this target sequence should be considered for normalization, which forms the above maximum a posteriori formula.

[0055] The maximum a posteriori probability is approximated by the longest common subsequence length (LCS) of the two sequences. That is, the longest common substring of the detected missing sequence and the alternative sequence is calculated. Unlike the calculation of the general longest common substring, multiple placeholders can be inserted into the missing sequence. The insertion condition of the placeholder is: for the detected scale value, if it does not appear in the alternative scale value sequence, it is replaced by the placeholder. The placeholder can match any number. For example: the matching degree of 0, 1, 7, 3, 4, 5 and 0, 1, 2, 3, 4, 5 is 6, and the matching sequence is 0, 1, 2, 3, 4, 5. Instead of 0, 1, 3, 4, 5 obtained by the ordinary longest common subsequence length (LCS). Because 7 does not appear in the matching scale sequence 0, 1, 2, 3, 4, 5, it is converted into a placeholder during matching and can be matched with 2. This effectively solves the problem of OCR recognizing digital errors. The final approximate maximum a posteriori probability is calculated by normalizing the LCS result with the length of the sequence of alternative scale values.

[0056] In the above embodiment, the longest common subsequence length takes into account the problems of missed detection and multiple detection, while the placeholder solves the problem of wrong detection. The LCS matching problem with placeholders is implemented through a dynamic programming algorithm.

[0057] By rotating the coordinate frame to detect the pointer and scale, the angle information of the scale and pointer can be obtained, but the angle value has a large error, and the precise angle value needs to be calculated. For the pointer on the dial, the distance between the pointer tip and the pointer center is longer than the tail end of the pointer. Therefore, the two points farthest from the center point of the pointer among the four vertices of the pointer coordinate frame are the pointer tip position, and the center value of the two vertices is the specific coordinate of the pointer tip. Then, the precise angle of the pointer can be calculated using the coordinates of the pointer tip and the coordinates of the pointer center, specifically,. Similarly, the precise angle of all scales relative to the pointer can be obtained. Then, the angle between the pointer and the adjacent scales on both sides can be used to convert the meter reading. For example: the angle of scale K1 is A1, the scale value is V1, the angle of scale K2 is A2, the scale value is V2, and the angle of the pointer between scales K1 and K2 is Ap. The calculation formula for the meter reading Vp is as follows:

[0058]

[0059] It should be noted that this embodiment uses the detected rotating frame to quickly calculate the pointer center, avoiding image correction and improving the generalization of the algorithm; uses a bipartite graph algorithm to match the OCR result with a known dial scale value sequence to obtain each scale value, effectively avoiding the reading calculation error problem caused by missed OCR characters and improving the robustness of the algorithm; uses the pointer tip, scale and pointer center to calculate the precise angle of the pointer and scale, effectively improving the reading accuracy. In summary, this method is a unified optimization of the reading recognition algorithms for a large number of different types of meters, and has strong robustness to shooting angles, image blur, lighting, meter dirt and aging. It is applicable to a variety of different meter recognition algorithms, while taking into account missed detection, multiple detection, and false detection problems, and can greatly improve the success rate of reading recognition.

[0060] Example 2

[0061] A computer program product involved in Example 2 of the present application is used to store a computer program. When the computer program runs on a computer, the method in any one of the implementation modes in Example 1 of the present application is implemented.

[0062] Example 3

[0063] A computer-readable storage medium according to Embodiment 3 of the present application, wherein the computer-readable storage medium stores a program code for execution by a device, wherein the program code includes steps for executing a method in any one of the implementations in Embodiment 1 of the present application;

[0064] Among them, the computer-readable storage medium can be a read-only memory (ROM), a static storage device, a dynamic storage device or a random access memory (RAM); the computer-readable storage medium can store program code, and when the program stored in the computer-readable storage medium is executed by the processor, the processor is used to execute the steps of the method in any one of the implementation methods in Example 1 of the present application.

[0065] Example 4

[0066] like Figure 5 As shown, an electronic device involved in Embodiment 4 of the present application includes a processor, a memory, and a program or instruction stored in the memory and executable on the processor, and when the program or instruction is executed by the processor, a method in any one of the implementations in Embodiment 1 of the present application is implemented;

[0067] Among them, the processor can adopt a general-purpose central processing unit (CPU), a microprocessor, an application specific integrated circuit (ASIC), a graphics processing unit (GPU) or one or more integrated circuits to execute relevant programs to implement the method in any one of the implementation methods in Example 1 of the present application.

[0068] The processor may also be an integrated circuit electronic device with signal processing capability. In the implementation process, each step of the method in any implementation of Embodiment 1 of the present application may be completed by an integrated logic circuit of hardware in the processor or by instructions in software form.

[0069] The above-mentioned processor may also be a general-purpose processor, a digital signal processor, an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application may be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in the embodiments of the present application may be directly embodied as being executed by a hardware decoding processor, or may be executed by a combination of hardware and software modules in a decoding processor. The software module may be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in a memory, and the processor reads the information in the memory, and completes the functions required to be performed by the unit included in the data processing device of the embodiment of the present application in combination with its hardware, or executes the method in any one of the implementation modes in the embodiment 1 of the present application.

[0070] The above are only preferred specific implementations of the present application; however, the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solution and its improved ideas of the present application within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for reading a universal pointer meter, characterized in that: include: Acquire a dial image, wherein the dial image includes scales and pointers; Performing rotation coordinate frame detection and OCR character recognition on the dial image for the scale and pointer to obtain the coordinates of the four vertices of the rotation coordinate frame, the position information of the pointer and the scale, and the scale value and the scale value position; Use bipartite graph algorithm to perform one-to-one matching between scales and scale values; Complete the scale value based on the preset scale value database; The average value of the coordinates of the intersection of the pointer and the extended line of the scale is taken as the coordinates of the pointer rotation center; Calculate the angle information of the scale and the pointer according to the center point coordinates of the two vertices farthest from the pointer selection center point in the coordinate frame and the pointer rotation center coordinates; The meter reading is calculated based on the angle between the pointer and the adjacent scales on both sides.

2. The method for reading a universal pointer meter according to claim 1, characterized in that: The acquisition of the dial image includes: The deep learning object detection model is trained through the annotated original dial image to obtain a dial detection model; The dial detection model is used to detect the image to be identified to accurately locate the position of the dial in the image to be identified, and then the dial image is extracted.

3. The method for reading a universal pointer meter according to claim 1, characterized in that: After OCR character recognition, the scale value of OCR character recognition is subjected to fault tolerance processing, including: Constructing a letter-number correspondence database, wherein the letter-number correspondence database contains letter and number combinations that are easily confused in OCR character recognition, It is determined whether the scale value is a letter. If the determination result is yes, the letter is corrected to a corresponding number in a letter-number correspondence database.

4. The method for reading a universal pointer meter according to claim 1, characterized in that: A one-to-one matching of scales and scale values ​​is performed using a bipartite graph algorithm, including: associating nearby scale values ​​for each scale, and completing the one-to-one matching of scales and scale values ​​by minimizing the total distance after matching, wherein the total distance is the sum of the Euclidean distances between all scales and corresponding scale values.

5. The method for reading a universal pointer meter according to claim 1 or 3, characterized in that: Complete the scale value based on the preset scale value database, including: Building a scale value database according to the business scenario of the meter, wherein the scale value database includes a known scale value sequence of the meter in the business scenario; According to the longest common subsequence length of the first sequence of scale values ​​and the second sequence in the scale value database, a complete scale value sequence is matched from the scale value database to complete the scale value completion.

6. The method for reading a universal pointer meter according to claim 5, characterized in that: According to the longest common subsequence length between the first sequence composed of scale values ​​and the second sequence in the scale value database, the first sequence is compared with the second sequence. If the first sequence contains a scale value that does not appear in the second sequence, the scale value is replaced with a placeholder, wherein the placeholder can match any number.

7. The method for reading a universal pointer meter according to claim 1, characterized in that: The calculation formula for the meter reading is: Among them, V p The angle of scale K1 is A1, and the scale value is V1; the angle of scale K2 is A2, and the scale value is V2. The angle of the pointer between scales K1 and K2 is A p .

8. A computer program product, characterized in that The computer program product stores computer instructions, and when the computer instructions are executed by a processor, the method according to any one of claims 1 to 5 is implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program code for execution by a device, wherein the program code includes steps for executing the method according to any one of claims 1 to 5.

10. An electronic device, characterized in that: The electronic device includes a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction implements the method according to any one of claims 1 to 5 when executed by the processor.

Citation Information

Patent Citations

  • Automobile driving seat instrument area multi-pointer type instrument automatic identification system and automobile driving seat instrument area multi-pointer type instrument automatic identification method based on template matching

    CN106960207A

  • Character detection method and device, detection equipment and storage medium

    CN111767908A

  • Obstacle detection evaluation method and device, equipment and storage medium

    CN112581527A

  • Certificate character recognition method and device, computer and storage medium

    CN114359923A

  • Pointer type instrument automatic reading method for inspection robot

    CN117152727A

Cited By

  • Multi-mode-based intelligent reading method for single-pointer meter of transformer substation

    CN119649352A

  • Reading identification method and system of square pointer meter

    CN120894770A

  • A reading recognition method and system for a square pointer watch

    CN120894770B

  • Meter reading method, device and system based on combination of electric power scene and large model

    CN121170809A

  • A meter reading method, device and system combining a power scenario with a large model

    CN121170809B