Pointer type instrument reading identification method based on key feature matching

Abstract

The invention provides a pointer type instrument reading recognition method based on key feature matching. The pointer type instrument reading recognition method comprises the steps that an instrument image database with a dial and pointer feature map is constructed; constructing a feature matching algorithm model based on the instrument image database; generating a matching image based on the feature matching algorithm model, and calculating coordinate data of the matching image; calculating a pointer deflection angle based on the coordinate data; and calculating a pointer reading based on the pointer deflection angle to realize reading identification of the pointer type instrument. According to the invention, the problems of relatively low efficiency and more man-machine interaction when the reading of the pointer type instrument is detected can be solved.

Description

Technical Field

[0001] This invention relates to the fields of machine vision and industrial inspection technology, and in particular to a method for recognizing pointer instrument readings based on key feature matching. Background Technology

[0002] Industrial instruments are generally divided into two main categories: pointer instruments and digital instruments. Pointer instruments typically lack a digital display interface, and the acquisition and entry of monitored data still require manual observation and recording. This manual interpretation method is labor-intensive and has low accuracy. In large-scale instrument scenarios in industrial fields or in toxic or hazardous environments, the drawbacks of manual instrument inspection and reading are even more pronounced.

[0003] Applying the latest research results in computer vision and artificial intelligence to the detection and reading recognition of pointer instruments is currently a mainstream research direction in industrial inspection, effectively solving problems such as accuracy fluctuations and weak universality. However, some issues also exist. On the one hand, in order to effectively detect and extract key targets such as dials, scale lines, scale numbers, and pointers, and to achieve high accuracy in instrument readings, it is usually necessary to perform contour marking on the target information contained in all images in the dataset, which is relatively inefficient. On the other hand, pointer instrument reading methods that combine target detection mostly utilize prior range information to complete the instrument reading. Information such as instrument type, instrument extreme values, and corresponding angles are stored in a database in advance, and the current reading can be obtained by associating the pointer position with the relevant prior information. The acquisition and recording of prior range information requires a lot of manual interaction, which does not conform to the design principles of automated instrument reading methods and also reduces the universality of the method.

[0004] Therefore, there is an urgent need for new detection methods to enable artificial intelligence-based visual inspection technology to detect the readings of pointer instruments in certain specific industrial environments. Summary of the Invention

[0005] This invention provides a pointer instrument reading recognition method based on key feature matching, which solves the problems of low efficiency and excessive manual interaction when detecting pointer instrument readings.

[0006] The first aspect of this invention provides a method for recognizing pointer-type instrument readings based on key feature matching, comprising the following steps:

[0007] Construct an instrument image database with feature maps of the dial and pointers;

[0008] A feature matching algorithm model is constructed based on the instrument image database;

[0009] A matching image is generated based on the feature matching algorithm model, and the coordinate data of the matching image is calculated;

[0010] Calculate the pointer deflection angle based on the coordinate data;

[0011] The pointer reading is calculated based on the pointer deflection angle, thereby enabling the recognition of readings on pointer-type instruments.

[0012] Furthermore, the construction of the instrument image database with dial and pointer feature maps includes the following steps:

[0013] The dial image data includes the dial type, the dial feature map, the dial feature map of the initial 0 position, the dial range, and the dial effective deflection angle.

[0014] Store pointer image data, which includes a feature map of the pointer, a feature map of the pointer axis, and a feature map of the pointer tip.

[0015] An instrument image database is constructed based on the dial image data and the pointer image data.

[0016] Furthermore, the construction of the feature matching algorithm model based on the instrument image database includes the following steps:

[0017] The feature images in the instrument image database are each made into templates to form feature template images;

[0018] The feature matching results are calculated using the Euclidean distance formula.

[0019] Furthermore, the process of generating a matching image based on the feature matching algorithm model and calculating the coordinate data of the matching image includes the following steps:

[0020] The instrument image of the pointer-type instrument to be detected is read into the feature matching algorithm model;

[0021] Calculate the similarity between the instrument image and the feature map in the instrument image database, and generate the corresponding dial image;

[0022] Using the geometric center of the dial image as the key point for positioning, the coordinates of the pointer axis, the dial zero point, and the pointer tip point are obtained.

[0023] Furthermore, the method of using the geometric center of the dial image as a key point for positioning and obtaining the coordinates of the pointer axis point, the dial zero point, and the pointer tip point includes the following steps:

[0024] S10. Use the feature map at the initial position of 0 to match on the dial image to generate a matching image. At the same time, use the geometric center of the dial image as a key point for positioning to obtain the coordinates of the zero point of the dial.

[0025] S20. Use the pointer feature map to match the matching image generated in step S10 to generate a pointer image;

[0026] S30. Match the pointer image generated in step S20 with the pointer axis feature map and the pointer tip feature map to generate the pointer axis image and the pointer tip image. At the same time, use the geometric center of the dial image as the key point for positioning to obtain the coordinates of the pointer axis point and the pointer tip point.

[0027] Furthermore, the calculation of the pointer deflection angle based on the coordinate data includes the following steps:

[0028] With the direction of pointer rotation as the positive direction, the pointer pivot point pointing to the zero point of the dial as the reference vector, and the pointer pivot point pointing to the pointer tip as the rotation vector;

[0029] The pointer deflection angle is calculated using the reference vector and the rotation vector.

[0030] Furthermore, the process of calculating the pointer reading based on the pointer deflection angle to realize the reading recognition of the pointer-type instrument includes the following steps:

[0031] Set up a reference data dictionary to associate the dial type with the dial range angle and the effective reading range;

[0032] Based on the feature matching algorithm model, the corresponding dial type image is generated. The dial range angle and effective reading range of the corresponding dial are obtained by using the dial type and the reference data dictionary.

[0033] Based on the pointer deflection angle, dial range angle, and effective reading range, the pointer reading is calculated using the following formula:

[0034]

[0035] In the formula, α represents the dial range angle, w represents the effective reading range, and θ is the pointer deflection angle.

[0036] A second aspect of the present invention provides a pointer-type instrument reading recognition system based on key feature matching, comprising:

[0037] The image database construction module is used to collect sample images of pointer-type instruments, extract dial and pointer features from the sample images, and construct an instrument image database containing dial and pointer features.

[0038] The model building module is used to call the feature data in the instrument image database, train the feature matching algorithm model through feature samples, and use the feature matching algorithm model to perform feature matching processing on the instrument image to be identified to generate a matching image.

[0039] The calculation module is used to calculate the coordinate data of the pointer key points in the matching image and the pointer deflection angle based on the preset image coordinate calculation rules;

[0040] The reading recognition module is used to pre-store the correspondence table between dial type, dial range angle and effective reading range, call the corresponding table and perform interpolation calculation in combination with pointer deflection angle, and output the final reading of the pointer instrument.

[0041] A third aspect of the present invention provides a computer device, comprising:

[0042] Memory, transceiver, processor, and bus system;

[0043] The memory is used to store programs;

[0044] The processor is configured to execute a program in the memory, including performing the method as described in any one of claims 1 to 7;

[0045] The bus system is used to connect the memory and the processor to enable communication between the memory and the processor.

[0046] A fourth aspect of the present invention provides a readable storage medium storing computer-readable instructions that, when executed by a processor, implement the steps of the method described above.

[0047] Compared with existing technologies, this invention has the following advantages: This invention constructs an instrument image database containing feature maps of multiple dial types and corresponding pointers, serving as a matching template. During feature matching, the instrument image to be read on-site is compared with the database template. Matching is achieved through similar feature points, eliminating the need for image-by-image annotation; only pre-processing of the standard template is required, and feature extraction is fully automated. In the reading detection stage, after matching, the corresponding dial and pointer images of the on-site instrument are generated, their coordinate data is calculated, and the pointer deflection angle is obtained through a vector angle algorithm. Combined with the dial type, the actual reading is calculated. This method eliminates the cumbersome outline marking of dials, scales, and pointers, completing the calculation only using feature point position information. Due to minimal human interaction, it significantly improves the efficiency of instrument reading.

[0048] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from an examination of the following, or may be learned from the practice of the invention. Attached Figure Description

[0049] Figure 1 A flowchart illustrating a pointer-type instrument reading recognition method based on key feature matching, provided in Embodiment 1 of the present invention;

[0050] Figure 2 This is a flowchart of the binarization operation provided in Embodiment 1 of the present invention;

[0051] Figure 3 This is a preprocessed image of the pointer-type instrument graph provided in Embodiment 2 of the present invention;

[0052] Figure 4 The dial image generated by matching is provided in Embodiment 2 of the present invention;

[0053] Figure 5 The pointer image generated by matching is provided in Embodiment 2 of the present invention. Detailed Implementation

[0054] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “corresponding to,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0055] Example 1

[0056] The implementation method in this embodiment can be implemented in a system, on a server, or on a terminal; no specific limitation is made. The method in this application will be described below from the perspective of system implementation. Please refer to... Figure 1 The method provided in this application includes the following steps:

[0057] Construct an instrument image database with feature maps of the dial and pointers.

[0058] A feature matching algorithm model is constructed based on the instrument image database.

[0059] A matching image is generated based on the feature matching algorithm model, and the coordinate data of the matching image is calculated.

[0060] The pointer deflection angle is calculated based on the coordinate data.

[0061] The pointer reading is calculated based on the pointer deflection angle, thereby enabling the recognition of readings on pointer-type instruments.

[0062] The method provided in Embodiment 1 first constructs an instrument image database containing dial feature maps and corresponding pointer feature maps. The instrument image database includes multiple dial types and their corresponding dial and pointer feature maps. These are used as instrument template images.

[0063] During feature matching, the field instrument image to be read is compared with the template image in the database to find similar feature points. There is no need to perform detailed target annotation for each input image. Only a limited number of standard templates need to be processed during the model building stage, and the feature extraction process is automated.

[0064] When detecting the reading of an instrument image, the field instrument image to be read generates corresponding dial and pointer images through feature matching, and calculates the coordinate data of each currently generated matching image. The pointer deflection angle is obtained by calculating the vector angle through the coordinate data, and the actual pointer reading of the pointer instrument is calculated based on the pointer deflection angle of the matching image.

[0065] The above-mentioned automated instrument reading method avoids the tedious work of marking the dial, scale lines, scale numbers, and pointer outlines. It only needs to use the position information of the pointer and feature points on the dial, and calculate the pointer deflection angle by calculating the vector angle. Based on the dial type detected by feature matching, the actual reading of the pointer instrument is calculated. It is highly efficient and requires less human interaction.

[0066] Furthermore, the construction of the instrument image database with dial and pointer feature maps includes the following steps: storing dial image data, which includes the dial type, dial feature map, dial range, effective dial deflection angle, and feature map of the dial's initial 0 position; storing pointer image data, which includes the pointer feature map, pointer axis feature map, and pointer tip feature map; and constructing the instrument image database based on the dial image data and the pointer image data.

[0067] The feature matching algorithm model constructed based on the instrument image database includes the following steps: making the feature images in the instrument image database into templates to form feature template images; and calculating the feature matching results using the Euclidean distance formula.

[0068] The feature templates are known and labeled. During feature matching, unknown samples are matched with all templates to calculate similarity.

[0069] When constructing the feature matching algorithm model, the first step is to create templates. Feature maps are converted into templates, and a function is defined to flatten the two-dimensional array image into a one-dimensional array. These templates are then stored in a list. Since we need to recognize various objects such as the dial and pointers, separate templates are required for each. For example, we might create a template for the dial's feature map, a template for the dial's range feature map, a template for the dial's effective deflection angle feature map, a template for the dial's initial 0 position feature map, a template for the pointer's feature map, a template for the pointer's axis feature map, and a template for the pointer's tip feature map, etc.

[0070] When calculating similarity, Euclidean distance is used. A function is defined to calculate Euclidean distance, as shown in the following formula:

[0071]

[0072] x i ,y i Let and be the coordinates of two points on the plane. During feature matching, the Euclidean distance between the image to be identified and the model is calculated. The model with the smallest distance is taken as the recognition result, and a matching image is generated.

[0073] Furthermore, the process of generating a matching image based on the feature matching algorithm model and calculating the coordinate data of the matching image includes the following steps: reading the instrument image of the pointer instrument to be detected into the feature matching algorithm model; calculating the similarity between the instrument image and the feature map in the instrument image database to generate the corresponding dial image; using the geometric center of the dial image as the key point for positioning, and obtaining the coordinates of the pointer axis point, the dial zero point, and the pointer tip point.

[0074] In the specific implementation plan, the image is first read; then the image data is filtered to eliminate noise interference and improve the accuracy of subsequent contour extraction; the image is converted to grayscale to simplify subsequent binarization and convolution operations, while preserving the brightness differences of the image to distinguish the dial, pointer, and background; binarization is performed to distinguish foreground and background image information; noise reduction is performed; contour fitting and key point coordinate extraction are then performed. After the previous processing, two main white connected regions, the dial contour and the pointer contour, are obtained in the image. Three key points are accurately extracted from these two contours: the pointer axis point, the dial zero point, and the pointer tip point.

[0075] The binarization operation is as follows: Figure 2 As shown, the image is first converted into a binary image through global binarization, and then the edges are judged based on the threshold. The parts within the threshold range are extracted and denoised to obtain the foreground image information; the parts outside the threshold range are used as the background image information.

[0076] The purpose of binarization is to simplify images, converting complex electronic images into binary images with only black and white values. This can quickly distinguish the foreground and background of an image, reduce redundant information, and at the same time reduce noise to make key information clearer, making it easier to identify and extract targets in the image, such as instrument pointers and object outlines.

[0077] Furthermore, the method of using the geometric center of the dial image as a key point for positioning and obtaining the coordinates of the pointer axis point, the dial zero point, and the pointer tip point includes the following steps:

[0078] S10. Use the feature map at the initial position of 0 to match the dial image to generate a matching image. At the same time, use the geometric center of the dial image as a key point for positioning to obtain the coordinates of the zero point of the dial.

[0079] S20. Use the pointer feature map to match the matching image generated in step S10 to generate a pointer image.

[0080] S30. Match the pointer image generated in step S20 with the pointer axis feature map and the pointer tip feature map to generate the pointer axis image and the pointer tip image. At the same time, use the geometric center of the dial image as the key point for positioning to obtain the coordinates of the pointer axis point and the pointer tip point.

[0081] First, after matching the instrument image with the dial feature map in the instrument image database, a dial image is generated. Based on the dial image, the information of the dial is obtained, including the dial type, the dial range of the corresponding dial type, and the effective deflection angle of the dial.

[0082] Then, the key points of the feature object are located. The geometric center of the generated dial image after matching is used as the key point. The feature map of the initial position of the dial is used to perform feature matching on the generated dial image and generate the result. The zero point coordinates (z1, z2) of the dial are then extracted by fitting the image contour.

[0083] The pointer image is generated by matching the image obtained from the previous step (z1, z2) with the pointer feature map template. The generated pointer image cannot yet be used to determine the coordinates of the pointer pivot point and pointer tip point; feature matching is required again. The pointer pivot point feature map and pointer tip point feature map are used to match the pointer image obtained in the previous step to generate the pointer pivot point image and pointer tip point image. Then, the pointer pivot point coordinates (c1, c2) and pointer tip point coordinates (p1, p2) are extracted through image contour fitting.

[0084] Furthermore, the calculation of the pointer deflection angle based on the coordinate data includes the following steps:

[0085] The pointer deflection angle is calculated using the pointer rotation direction as the positive direction, the pointer axis pointing to the zero point of the dial as the reference vector, and the pointer axis pointing to the pointer tip as the rotation vector.

[0086] The instrument diagram of the pointer-type instrument to be read is matched with the instrument image database. The dial zero point (z1, z2), pointer axis point (c1, c2), and pointer tip point (p1, p2) are extracted through image contour fitting. Taking the pointer rotation direction as the positive direction, the pointer axis point (c1, c2) pointing to the dial zero point (z1, z2) is used as the reference vector, and the pointer axis point (c1, c2) pointing to the pointer tip point (p1, p2) is used as the rotation vector. That is, the reference vector v1 = (z1 - c1, z2 - c2), and the rotation vector v2 = (p1 - c1, p2 - c2). These are calculated using the following formulas:

[0087]

[0088] Calculate the pointer deflection angle, where θ is the pointer deflection angle. Substitute vector v1 into vector a in the formula and vector v2 into vector b in the formula to calculate the value of θ.

[0089] Furthermore, the method of calculating the pointer reading based on the pointer deflection angle to achieve reading recognition of the pointer-type instrument includes the following steps: setting up a reference data dictionary to associate the dial type with the dial range angle and the effective reading range; generating a corresponding dial type image based on a feature matching algorithm model, and obtaining the dial range angle and the effective reading range of the corresponding dial through the dial type and the reference data dictionary; calculating the pointer reading based on the pointer deflection angle, the dial range angle, and the effective reading range using the following formula:

[0090]

[0091] In the formula, α represents the dial range angle, w represents the effective reading range, and θ is the pointer deflection angle.

[0092] This invention also provides a pointer-type instrument reading recognition system based on key feature matching, comprising:

[0093] The image database construction module is used to collect sample images of pointer-type instruments, extract dial and pointer features from the sample images, and construct an instrument image database containing dial and pointer features.

[0094] The model building module is used to call the feature data in the instrument image database, train the feature matching algorithm model through feature samples, and use the feature matching algorithm model to perform feature matching processing on the instrument image to be identified to generate a matching image.

[0095] The calculation module is used to calculate the coordinate data of the pointer key points in the matched image and the pointer deflection angle based on the preset image coordinate calculation rules.

[0096] The reading recognition module is used to pre-store the correspondence table between dial type, dial range angle and effective reading range, call the corresponding table and perform interpolation calculation in combination with pointer deflection angle, and output the final reading of the pointer instrument.

[0097] This invention also provides a computer device, including: a memory, a transceiver, a processor, and a bus system; wherein the memory is used to store a program; the processor is used to execute the program in the memory, including executing the method described above; and the bus system is used to connect the memory and the processor to enable communication between the memory and the processor.

[0098] This invention also provides a readable storage medium storing computer-readable instructions, which, when executed by a processor, implement the steps of the method described above.

[0099] Example 2

[0100] Embodiment 2 of the present invention provides a method for recognizing pointer-type instrument readings based on key feature matching, as detailed below:

[0101] Using the pointer instrument reading recognition system, computer equipment, and readable storage medium provided by this invention, the image of the pointer instrument to be read is input into the system, and the system reads the image.

[0102] First, the imported image (an image of a pointer-type instrument obtained by an industrial camera) is preprocessed to generate a result like this. Figure 3 The original image shown shows a reading of approximately 16. Using the dial feature map... Figure 3 Matching is performed on the original image shown to generate a result like... Figure 4 The dial image shown is used to obtain the corresponding feature template image. The range angle is α = 360°, and the effective reading range is w = 100°.

[0103] The feature map with the initial position of 0 is matched on the generated image, and the corresponding dial zero point coordinates are extracted by key point localization after image contour fitting (120, 28).

[0104] The pointer feature map template is used to match the results generated in the previous step, resulting in, for example: Figure 5 The pointer image shown is then used to match the pointer image generated in the previous step using the axis feature map and the pointer tip feature map respectively. After image contour fitting and key point localization, the corresponding pointer axis coordinates (117, 118) and pointer tip coordinates (213, 62) are extracted.

[0105] Calculate the reference vector v1 = (6, -90) and the rotation vector v2 = (96, -56) using the obtained coordinates, and then use the formula:

[0106]

[0107] The calculated deflection angle θ = 55.667 was obtained using the formula. Converting the deflection angle θ, the reading r = 15.466 is obtained. r = 15.466 is the system's reading from the photographed pointer instrument image, and 16 is the actual reading of the pointer instrument. The results show that the error for the dial reading is within ±0.73, and the measurement error for the angle of this type of instrument is within ±2.5, which can meet the actual requirements of the corresponding industrial site.

[0108] This invention can be widely applied to reading tasks of pointer instruments in harsh working environments such as confined spaces like pipes, high-temperature workshops, high electromagnetic radiation, and high dust, protecting the safety and health of measurement personnel while avoiding large measurement errors caused by subjective factors of the measurement personnel.

[0109] This invention avoids the tedious work of marking the dial, scale lines, scale numbers, and pointer outlines. It only needs to use the position information of three key feature points on the pointer and dial to calculate the pointer deflection angle by calculating the vector angle. Based on the detected dial type, the actual reading of the pointer instrument can be calculated.

[0110] Those skilled in the art will recognize that the units of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of the invention.

[0111] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical functional division. In actual implementation, there may be other division methods, such as multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored.

[0112] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0113] If the integrated unit is implemented as 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 invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0114] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for recognizing pointer-type instrument readings based on key feature matching, characterized in that, Includes the following steps: Construct an instrument image database with feature maps of the dial and pointers; A feature matching algorithm model is constructed based on the instrument image database; A matching image is generated based on the feature matching algorithm model, and the coordinate data of the matching image is calculated; Calculate the pointer deflection angle based on the coordinate data; The pointer reading is calculated based on the pointer deflection angle, thereby enabling the recognition of readings on pointer-type instruments.

2. The pointer instrument reading recognition method based on key feature matching according to claim 1, characterized in that, The construction of the instrument image database with dial and pointer feature maps includes the following steps: The dial image data includes the dial type, the dial feature map, the dial feature map of the initial 0 position, the dial range, and the dial effective deflection angle. Store pointer image data, which includes a feature map of the pointer, a feature map of the pointer axis, and a feature map of the pointer tip. An instrument image database is constructed based on the dial image data and the pointer image data.

3. The pointer instrument reading recognition method based on key feature matching according to claim 1, characterized in that, The feature matching algorithm model constructed based on the instrument image database includes the following steps: The feature images in the instrument image database are each made into templates to form feature template images; The feature matching results are calculated using the Euclidean distance formula.

4. The pointer instrument reading recognition method based on key feature matching according to claim 1, characterized in that, The process of generating a matching image based on the feature matching algorithm model and calculating the coordinate data of the matching image includes the following steps: The instrument image of the pointer-type instrument to be detected is read into the feature matching algorithm model; Calculate the similarity between the instrument image and the feature map in the instrument image database, and generate the corresponding dial image; Using the geometric center of the dial image as the key point for positioning, the coordinates of the pointer axis, the dial zero point, and the pointer tip point are obtained.

5. The pointer instrument reading recognition method based on key feature matching according to claim 4, characterized in that, The method of using the geometric center of the dial image as a key point for positioning and obtaining the coordinates of the pointer axis point, the dial zero point, and the pointer tip point includes the following steps: S10. Use the feature map at the initial position of 0 to match on the dial image to generate a matching image. At the same time, use the geometric center of the dial image as a key point for positioning to obtain the coordinates of the zero point of the dial. S20. Use the pointer feature map to match the matching image generated in step S10 to generate a pointer image; S30. Match the pointer image generated in step S20 with the pointer axis feature map and the pointer tip feature map to generate the pointer axis image and the pointer tip image. At the same time, use the geometric center of the dial image as the key point for positioning to obtain the coordinates of the pointer axis point and the pointer tip point.

6. The pointer instrument reading recognition method based on key feature matching according to claim 1, characterized in that, The calculation of the pointer deflection angle based on the coordinate data includes the following steps: With the direction of pointer rotation as the positive direction, the pointer pivot point pointing to the zero point of the dial as the reference vector, and the pointer pivot point pointing to the pointer tip as the rotation vector; The pointer deflection angle is calculated using the reference vector and the rotation vector.

7. The pointer instrument reading recognition method based on key feature matching according to claim 1, characterized in that, The process of calculating the pointer reading based on the pointer deflection angle to achieve reading recognition of pointer-type instruments includes the following steps: Set up a reference data dictionary to associate the dial type with the dial range angle and the effective reading range; Based on the feature matching algorithm model, the corresponding dial type image is generated. The dial range angle and effective reading range of the corresponding dial are obtained by using the dial type and the reference data dictionary. Based on the pointer deflection angle, dial range angle, and effective reading range, the pointer reading is calculated using the following formula: In the formula, α represents the dial range angle, w represents the effective reading range, and θ is the pointer deflection angle.

8. A pointer-type instrument reading recognition system based on key feature matching, characterized in that, include: The image database construction module is used to collect sample images of pointer-type instruments, extract dial and pointer features from the sample images, and construct an instrument image database containing dial and pointer features. The model building module is used to call the feature data in the instrument image database, train the feature matching algorithm model through feature samples, and use the feature matching algorithm model to perform feature matching processing on the instrument image to be identified to generate a matching image. The calculation module is used to calculate the coordinate data of the pointer key points in the matching image and the pointer deflection angle based on the preset image coordinate calculation rules; The reading recognition module is used to pre-store the correspondence table between dial type, dial range angle and effective reading range, call the corresponding table and perform interpolation calculation in combination with pointer deflection angle, and output the final reading of the pointer instrument.

9. A computer device, characterized in that, include: Memory, transceiver, processor, and bus system; The memory is used to store programs; The processor is configured to execute a program in the memory, including performing the method as described in any one of claims 1 to 7; The bus system is used to connect the memory and the processor to enable communication between the memory and the processor.

10. A readable storage medium storing computer-readable instructions, characterized in that, When the computer-readable instructions are executed by a processor, they implement the steps of the method as described in any one of claims 1 to 7.

Images