A full-range text information recognition method and system for wheel pair signboards

Abstract

This invention relates to the field of image recognition and intelligent detection, and discloses a method and system for omnidirectional text information recognition of wheelset identification signs. The method includes: issuing a synchronous acquisition command in response to detecting that a train wheelset has arrived at a preset detection area; synchronously adjusting the lighting environment of the identification signs on both sides of the train wheelset in response to the synchronous acquisition command, and performing image acquisition to obtain two identification sign images; for each identification sign image, extracting the corresponding text region image based on the text region, and performing optical character recognition on the text region image to obtain the corresponding recognition result. It can accurately recognize text on identification signs tilted at any angle, without requiring manual adjustment of the wheelset or identification sign position, and is adaptable to various wheelset installation deviation scenarios; effectively suppressing problems such as reflection, shadows, and uneven lighting; and enabling simultaneous acquisition and recognition from both sides, achieving "recognition completed as soon as the wheelset passes," reducing labor costs and management loopholes.

Description

Technical Field

[0001] This disclosure relates to the field of image recognition and intelligent detection, and in particular to a method for recognizing omnidirectional text information on wheelset signs, a system for recognizing omnidirectional text information on wheelset signs, a storage medium, an electronic device, and a computer program product. Background Technology

[0002] Wheelsets are core load-bearing components of rail transit vehicles, directly impacting operational safety and requiring strict adherence to service life and inspection cycle requirements. Each wheelset is equipped with a unique identification plate carrying key data such as wheelset number, production information, and maintenance records. This plate is the core carrier for tracking wheelset service life and mileage, and a crucial basis for ensuring timely maintenance and compliant scrapping. With the informatization and intelligent upgrading of the rail transit industry, the demand for efficient and accurate information collection in wheelset management is increasingly urgent, as traditional identification methods are no longer adequate for the requirements of large-scale, standardized operation and maintenance management.

[0003] Currently, traditional identification methods include manual visual inspection and traditional machine vision inspection. Among them: (1) Manual visual inspection technology: Core tools: No dedicated equipment is available; the system relies on visual observation by operators, supplemented by paper and pen or handheld terminals for recording.

[0004] Implementation process: After the vehicle comes to a complete stop, the operator approaches the wheelset identification plate, reads the text information visually, and manually enters it into the system or records it on a paper document; if the identification plate is stained or tilted, it needs to be cleaned manually or the viewing angle needs to be adjusted.

[0005] Application scenarios: minor maintenance, emergency repairs, and low-frequency wheel assembly inspections.

[0006] (2) Traditional machine vision inspection technology: Core equipment: a single industrial camera, a standard supplementary lighting device, and a local data processing terminal.

[0007] Implementation process: An industrial camera is fixedly installed at the maintenance station. After the wheelset enters the inspection area, the camera is manually triggered or triggered by a simple sensor to take a picture. The acquired image is transmitted to the terminal. After preprocessing such as noise reduction and contrast adjustment, the text information is extracted by traditional OCR algorithm. If the recognition fails, the wheelset position or camera parameters need to be manually adjusted and the picture taken again.

[0008] Application scenarios: medium and large-sized maintenance bases, end-of-line inspection of wheelset production lines, suitable for scenarios where the position of the signboard is relatively fixed and the text direction is regular.

[0009] However, the current traditional identification methods have the following defects: (1) For manual visual inspection technology: it is inefficient, costly and has large errors. Its main defects are: a single person can only identify 80-100 wheelsets per day, which is far less efficient than the needs of large-scale management; manual recording is prone to missed detection and false detection due to fatigue and subjective judgment differences, with an error rate of 5%-8%; it relies on a large number of manpower for a long time, the operation and maintenance costs are high, and the data is difficult to trace, which cannot meet the requirements of standardized management. (2) For traditional machine vision inspection technology: it has poor adaptability and limited recognition accuracy. Its main defects are: the text direction adaptability is weak, and when the tilt angle of the sign exceeds 15, the recognition success rate drops sharply to below 60%; the supplementary lighting scheme is simple and cannot cope with the problems of surface reflection of the sign and uneven lighting in the workshop, resulting in poor image quality and affecting the recognition effect; it can only collect images in one direction and cannot cover the signs on both sides of the wheelset at the same time, requiring multiple shots, which is cumbersome. The root cause is that the design was not optimized for the characteristics of wheel signage, which is characterized by "variable posture and complex environment". Traditional OCR algorithms lack multi-directional text processing capabilities, and the layout of supplementary lighting and image acquisition is not coordinated, which makes it impossible to solve the imaging and recognition problems in complex scenarios.

[0010] Therefore, a new identification method is urgently needed to avoid the defects mentioned above. Summary of the Invention

[0011] The purpose of this invention is to provide at least one method and device for omnidirectional text information recognition of wheelset signs. By adopting an integrated design of 'double-sided symmetrical image acquisition layout + shadowless supplementary lighting optimization + deep learning OCR algorithm', it achieves automated and high-precision text recognition of wheelset signs in omnidirectional and complex environments, solves the problem of sensitivity of existing technologies to text direction and lighting conditions, and improves the level of intelligence and standardization of wheelset management.

[0012] To address the aforementioned technical problems, at least one embodiment of this application provides a method for omnidirectional text information recognition of wheelset identification signs, the method comprising: In response to the detection that the train wheelset has reached the preset detection area, a synchronous data acquisition command is issued; In response to the synchronous acquisition command, the lighting environment of the signs on both sides of the train wheelset is adjusted synchronously, and an image acquisition operation is performed to obtain two signs images; For each sign image, the corresponding text region image is extracted based on the text region, and optical character recognition is performed on the text region image to obtain the corresponding recognition result.

[0013] At least one embodiment of this application also provides an omnidirectional text information recognition system for wheelset signage, comprising: The image acquisition and control unit is used to issue a synchronous acquisition command in response to the detection that the train wheelset has reached the preset detection area; An image acquisition unit, communicatively connected to the image acquisition unit, is used to respond to the synchronous acquisition command to simultaneously perform image acquisition operations on both sides of the train wheelset for the signboard, thereby obtaining two signboard images; A data processing and analysis unit, which is communicatively connected to the data processing and analysis unit, is used to extract the corresponding text region image based on the text region of each sign image, and to perform optical character recognition on the text region image to obtain the corresponding recognition result.

[0014] At least one embodiment of this application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described above.

[0015] At least one embodiment of this application also provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method described above.

[0016] At least one embodiment of this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method described above.

[0017] This application provides a method, system, storage medium, electronic device, and computer program product for omnidirectional text information recognition of wheelie signs. Compared to existing technologies, this application achieves automated, high-precision text recognition of wheelie signs in complex environments through an integrated design of 'dual-sided symmetrical image acquisition layout + shadowless supplementary lighting optimization + deep learning OCR algorithm', solving the problem of sensitivity to text direction and lighting conditions in existing technologies, and improving the intelligence and standardization of wheelie management. More specific beneficial effects include: (1) Strong recognition adaptability, covering all aspects of text: It can accurately recognize the text on signs tilted at any angle from 0° to 360° without the need for manual adjustment of wheelsets or sign positions. Its adaptability to changes in sign posture is significantly better than traditional technologies, and it is suitable for various wheelset installation deviation scenarios. (2) Stable imaging quality and outstanding anti-interference ability: The dual-sided symmetrical acquisition and shadowless ring light source design effectively suppresses reflection, shadow and uneven lighting problems, improving image clarity and contrast by more than 40% in complex environments, providing a reliable guarantee for high-accuracy recognition. (3) High detection efficiency, achieving full automation: Simultaneous acquisition and recognition from both sides, with a single detection time of ≤300ms, and a single device can recognize 300-400 wheelsets per day, which is 3-4 times that of manual inspection and 1.5-2 times that of traditional machine vision inspection; no manual intervention is required throughout the process, achieving "recognition completed as soon as wheelsets pass by"; (4) Accurate and traceable data reduces management costs: The recognition error rate is ≤0.8%, far lower than the 5%-8% of manual inspection; the recognition results are linked with the back-end database in real time, automatically updating maintenance records, realizing the standardization of wheel pair management and data traceability, reducing labor costs and management loopholes.

[0018] In some optional embodiments, the omnidirectional text information recognition method for wheelset signage further includes: If the recognition result does not meet the preset conditions, a new synchronization acquisition command is issued to re-acquire the sign image and perform optical character recognition, thereby obtaining the recognition result again. This aims to ensure the accuracy of the recognition result as much as possible.

[0019] In some optional embodiments, the step of adjusting the lighting environment includes: Obtain the current ambient light intensity of the environment where the sign is located; Based on the correspondence table between the current ambient light intensity and the preset supplementary light amount, the target supplementary light amount is determined; Based on the target supplementary light amount, a supplementary light control command is generated to drive the supplementary light device to provide adaptive supplementary light to the train wheelsets. The shadowless ring light source automatically adjusts its brightness according to the ambient light intensity (for example, 500-800 lux in strong light environments and 1500-2000 lux in weak light environments), thereby suppressing reflections and shadows on the sign surface.

[0020] In some optional embodiments, obtaining the current ambient light intensity of the environment where the sign is located includes: The system acquires the current ambient light intensity of the area where the sign is located at preset time intervals. Based on these preset time intervals, it acquires the current ambient light intensity multiple times and adjusts the target supplementary lighting amount according to changes in ambient light. By performing multiple adjustments to the supplementary lighting amount, adaptive supplementary lighting for train wheelsets is achieved.

[0021] In some optional embodiments, the step of determining the identification result includes: The text region is obtained from the sign image using a deep learning-based text detection algorithm, and the tilt angle of the text region is calculated. The text area is rotated and corrected according to the tilt angle so that the text in the text area is in a horizontal direction; The corrected text area is recognized using OCR to obtain the recognition results. It can accurately recognize the text on signs tilted at any angle from 0° to 360° without the need for manual adjustment of the wheelset or sign position. Its adaptability to changes in sign posture is significantly better than traditional technologies, and it is suitable for various wheelset installation deviation scenarios.

[0022] In some optional embodiments, the recognition result includes text recognition result and corresponding confidence score; the method further includes: If the confidence level of each of the recognition results is not less than the confidence level threshold, retrieval information is generated based on the text recognition result corresponding to the manufacturing information sign; wherein, the sign includes manufacturing information signs and maintenance information signs; Based on the retrieved information and the preset wheelset file information, the cumulative operating mileage and usage cycle of the train wheelset are determined; Based on the accumulated mileage and usage cycle, it is determined whether the train wheelsets have reached the maintenance or scrapping threshold. If the threshold is reached, a health status reminder message is issued. By comparing the OCR recognition results with pre-stored wheelset information in the background database, the wheelset usage cycle and maintenance status are automatically calculated, thereby updating maintenance records in a timely manner and supporting historical traceability. Simultaneously, the health status reminder message allows operators to grasp the wheelset's health information immediately and make timely plans or adjustments to subsequent operations based on the current health information.

[0023] In some optional embodiments, the method further includes: The retrieval results, identification results for each sign, and current inspection time are stored in the database to update the train wheelset maintenance records. The identification results are linked to the backend database in real time, automatically updating maintenance records, thus standardizing wheelset management and ensuring data traceability, reducing labor costs and management loopholes.

[0024] In some optional embodiments, the recognition result includes text recognition result and corresponding confidence level; the system further includes: a data retrieval processing unit, a data retrieval unit, and a health status determination unit; wherein: The retrieval data processing unit is used to generate retrieval information based on the text recognition results of each recognition result, provided that the confidence level of each recognition result is not less than the confidence level threshold. The data retrieval unit is communicatively connected to the retrieval data processing unit and is used to determine the cumulative operating mileage and usage cycle of the train wheelset based on the retrieval information and the preset wheelset file information. The health status determination unit is communicatively connected to the data retrieval unit and is used to determine whether the train wheelset has reached the maintenance or scrapping threshold based on the cumulative operating mileage and the usage cycle, and to issue a health status reminder message if the maintenance or scrapping threshold is reached.

[0025] In some optional embodiments, the system further includes a light sensor, a supplementary lighting control unit, and a supplementary lighting device; wherein: The light sensor is used to obtain the current ambient light intensity of the environment in which the sign is located; The data processing module is communicatively connected to the light sensor and is also used to determine the target supplementary light amount based on the current ambient light intensity. The supplementary lighting control unit is communicatively connected to the data processing unit and is used to generate supplementary lighting control commands based on the target supplementary lighting amount to drive the supplementary lighting device to adaptively supplement the signboard. The supplementary lighting device is communicatively connected to the supplementary lighting control unit and is used to emit light to the train wheelsets to meet preset imaging requirements. Attached Figure Description

[0026] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0027] Figure 1 A flowchart illustrating a method for omnidirectional text information recognition of wheelset identification signs provided in this embodiment of the present disclosure; Figure 2 This is a flowchart of a method for monitoring the health status of train wheelsets, provided as an embodiment of this disclosure. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the various embodiments of the present invention will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details have been presented in the various embodiments of the present invention to enable the reader to better understand the present invention. However, the technical solutions claimed in the present invention can be implemented even without these technical details and various changes and modifications based on the following embodiments.

[0029] Example 1: The embodiments of the present invention relate to a method for omnidirectional text information recognition of wheelset signage.

[0030] The following is a detailed description of the implementation details of the omnidirectional text information recognition method for wheel-mounted signage in this embodiment. The following content is only for the convenience of understanding and is not necessary for implementing this solution.

[0031] The omnidirectional text information recognition method for wheelset identification plates in this embodiment can be applied to electronic devices with communication, computing, and data storage capabilities. For example... Figure 1 As shown, the omnidirectional text information recognition method for wheelset signage provided in this embodiment includes the following steps: Step 110: In response to the detection that the train wheelset has reached the preset detection area, a synchronous acquisition command is issued.

[0032] Optionally, wheelset signals can be detected using photoelectric sensors. Additionally, a preset detection area can be set according to specific circumstances.

[0033] Specifically, when the wheelset enters the preset detection area along the track, the photoelectric sensor detects the wheelset signal and immediately issues a synchronous acquisition command to ensure that images are acquired on both sides of the wheelset at the same time, avoiding information loss due to shooting time difference.

[0034] Step 120: In response to the synchronous acquisition command, synchronously adjust the lighting environment of the signs on both sides of the train wheelset and perform image acquisition to obtain two signs images.

[0035] Upon receiving the synchronous acquisition command, images of the identification plates on both sides of the wheelset can be captured synchronously using a high-resolution industrial camera. Each set of shots generates two clear images (corresponding to the identification plates on both sides of the wheelset, respectively).

[0036] In some alternative embodiments, the step of adjusting the lighting environment includes: Obtain the current ambient light intensity of the environment where the sign is located; Based on the correspondence table between the current ambient light intensity and the preset supplementary light amount, the target supplementary light amount is determined; Based on the target supplementary light amount, a supplementary light control command is generated to drive the supplementary light device to perform adaptive supplementary light on the train wheelsets.

[0037] Specifically, upon receiving the synchronous acquisition command, the brightness of the supplementary lighting device is automatically adjusted according to the ambient light intensity (e.g., 500-800 lux in strong light environment and 1500-2000 lux in weak light environment), thereby suppressing reflections and shadows on the surface of the sign.

[0038] In some possible cases, a correspondence table between ambient light intensity and supplementary light amount can be preset (i.e., a preset supplementary light amount correspondence table), and the target supplementary light amount can be determined from the correspondence table based on the ambient light intensity.

[0039] In some optional embodiments, obtaining the current ambient light intensity of the environment where the sign is located includes: The current ambient light intensity of the environment where the sign is located is obtained at preset time intervals.

[0040] The preset time interval can be 1 second, 1 minute, 10 minutes, 20 minutes, half an hour, or 1 hour, etc. Based on the preset time interval, the current ambient light intensity is acquired multiple times, and the target supplementary lighting amount is adjusted according to the changes in ambient light. By making multiple adjustments to the supplementary lighting amount, adaptive supplementary lighting for train wheelsets can be achieved.

[0041] Step 130: For each sign image, extract the corresponding text region image based on the text region, and perform optical character recognition on the text region image to obtain the corresponding recognition result.

[0042] In some embodiments, the step of determining the identification result includes: The text region is obtained from the sign image using a deep learning-based text detection algorithm, and the tilt angle of the text region is calculated. The text area is rotated and corrected according to the tilt angle so that the text in the text area is in a horizontal direction; The corrected text region is used to perform character recognition, and the recognition result is obtained.

[0043] After receiving the original sign image, preprocessing is first performed: Gaussian filtering algorithm is used to remove image noise, histogram equalization technology is used to enhance local contrast and eliminate the effects of uneven lighting; then, based on color and shape features, image segmentation algorithm is used to roughly locate the sign area, and then edge detection and morphological operations are combined to accurately locate the minimum bounding rectangle of the sign and remove background interference.

[0044] Then, a deep learning-based text detection algorithm is used to scan the text blocks within the sign area and calculate the main direction (tilt angle) of the text. The text area is rotated and corrected according to the tilt angle, turning text in any direction into a horizontal direction. Deep learning OCR is used to recognize characters in the corrected text area and output the recognition results. The recognition results include the text recognition results and the corresponding confidence scores.

[0045] In some embodiments, the method further includes: If the recognition result does not meet the preset conditions, a synchronization acquisition command is reissued to re-acquire the sign image and perform optical character recognition, thereby obtaining the recognition result again.

[0046] The preset condition can be set to the point where the confidence level of the device result is less than a confidence threshold. The confidence threshold can be set to 90%, and if the result is lower than this threshold, the test will be repeated.

[0047] The omnidirectional text information recognition method for wheel set signs provided in this embodiment, compared with existing technologies, can accurately recognize the text on signs tilted at any angle from 0° to 360° without manual adjustment of the wheel set or sign position. Its adaptability to changes in sign posture is significantly superior to traditional technologies, and it is suitable for various wheel set installation deviation scenarios. The dual-sided symmetrical acquisition and shadowless ring light source design effectively suppresses reflections, shadows, and uneven lighting problems, improving image clarity and contrast by more than 40% in complex environments, providing a reliable guarantee for high-accuracy recognition. Dual-sided synchronous acquisition and recognition, with a single detection time of ≤300ms, allows a single device to recognize 300-400 sets of wheelsets per day, which is 3-4 times that of manual inspection and 1.5-2 times that of traditional machine vision inspection. The entire process requires no manual intervention, achieving "recognition completed as soon as the wheel set passes." The recognition error rate is ≤0.8%, far lower than the 5%-8% of manual inspection. The recognition results are linked with the backend database in real time, automatically updating maintenance records, achieving standardization and data traceability for wheel set management, reducing labor costs and management loopholes.

[0048] Example 2: Based on the above embodiments, this embodiment further explains and illustrates how the omnidirectional text information recognition method for wheelset identification plates provided in the above embodiments achieves wheelset health status monitoring.

[0049] For reference Figure 2 , Figure 2 This disclosure provides a flowchart of a method for monitoring the health status of train wheelsets, which further enables the monitoring of the health status of wheelsets after obtaining the recognition results of the sign image.

[0050] In some embodiments, the aforementioned method for omnidirectional text information recognition of wheelset signage further includes: Provided that the confidence level of each recognition result is not less than the confidence level threshold, retrieval information is generated based on the text recognition result corresponding to the manufacturing information sign; wherein, the sign includes manufacturing information signs and maintenance information signs; Based on the retrieved information and the preset wheelset file information, the cumulative operating mileage and usage cycle of the train wheelset are determined; Based on the cumulative operating mileage and the usage cycle, it is determined whether the train wheelset has reached the maintenance or scrapping threshold, and if the maintenance or scrapping threshold is reached, a health status reminder message is issued.

[0051] If the cumulative mileage and usage period of a train wheelset cannot be retrieved from the preset wheelset archive information based on the search information, then the cumulative mileage and usage period of the wheelset can be determined by calculation based on the information retrieved from the preset wheelset archive information.

[0052] In some embodiments, the aforementioned method for omnidirectional text information recognition of wheelset signage further includes: The retrieval results of the search information, the identification results corresponding to each sign, and the current detection time are stored in the database to update the maintenance records of the train wheelsets.

[0053] Specifically, the search results can include successful and unsuccessful searches (where the cumulative mileage and usage period cannot be obtained through retrieval or calculation).

[0054] The data processing and analysis unit retrieves and matches the identification results with the pre-stored wheelset archive information in the background database, automatically calculates the cumulative running mileage and usage cycle of the wheelset, and determines whether it has reached the maintenance or scrapping threshold; it uploads the identification results, matching conclusions, detection time and other information to the background database, updates the wheelset maintenance records, and supports historical traceability and report generation; if the matching fails, it sends a reminder to the operation and maintenance terminal and waits for manual review.

[0055] In some embodiments, if the confidence level of the recognition result obtained based on the re-acquired sign image is still less than the confidence level threshold, a reminder can be sent to the maintenance terminal to wait for manual review.

[0056] The method provided in this embodiment enables the monitoring of wheelset health information based on the completed text information recognition. By comparing the OCR recognition results with pre-stored wheelset information in the background database, the wheelet's usage cycle and maintenance status are automatically calculated, thereby updating maintenance records in a timely manner and supporting historical traceability. Simultaneously, health status alert messages allow operators to grasp the wheelset's health information immediately and make timely plans or adjustments to subsequent operations based on the current health information of the wheelsets.

[0057] Example 3: Based on the above embodiments, this embodiment provides an omnidirectional text information recognition system for wheelset identification signs. The system specifically includes: The image acquisition and control unit is used to issue a synchronous acquisition command in response to the detection that the train wheelset has reached the preset detection area; An image acquisition unit, communicatively connected to the image acquisition unit, is used to respond to the synchronous acquisition command to simultaneously perform image acquisition operations on both sides of the train wheelset for the signboard, thereby obtaining two signboard images; A data processing and analysis unit, which is communicatively connected to the data processing and analysis unit, is used to extract the corresponding text region image based on the text region of each sign image, and to perform optical character recognition on the text region image to obtain the corresponding recognition result.

[0058] In some embodiments, the system further includes: a data retrieval processing unit, a data retrieval unit, and a health status determination unit; wherein: The retrieval data processing unit is used to generate retrieval information based on the text recognition results of each recognition result, provided that the confidence level of each recognition result is not less than the confidence level threshold. The data retrieval unit is communicatively connected to the retrieval data processing unit and is used to determine the cumulative operating mileage and usage cycle of the train wheelset based on the retrieval information and the preset wheelset file information. The health status determination unit is communicatively connected to the data retrieval unit and is used to determine whether the train wheelset has reached the maintenance or scrapping threshold based on the cumulative operating mileage and the usage cycle, and to issue a health status reminder message if the maintenance or scrapping threshold is reached.

[0059] In some embodiments, the system further includes a light sensor, a supplementary lighting control unit, and a supplementary lighting device; wherein: The light sensor is used to obtain the current ambient light intensity of the environment in which the sign is located; The data processing module is communicatively connected to the light sensor and is also used to determine the target supplementary light amount based on the current ambient light intensity. The supplementary lighting control unit is communicatively connected to the data processing unit and is used to generate supplementary lighting control commands based on the target supplementary lighting amount to drive the supplementary lighting device to adaptively supplement the signboard. The supplementary lighting device is communicatively connected to the supplementary lighting control unit and is used to emit light to the train wheelsets to meet preset imaging requirements.

[0060] As a specific example, based on the omnidirectional text information recognition system for wheelset signs provided in this embodiment, the following is a specific application example: It mainly includes an image acquisition and control unit, an image acquisition unit, and a data processing and analysis unit. The image acquisition units are symmetrically arranged on both sides of the track and connected to the data processing and analysis unit via a high-speed industrial communication interface to ensure real-time transmission of image data. Each image acquisition unit consists of a set of high-resolution industrial cameras and a matching shadowless ring light source. The light source design effectively suppresses reflections and shadows on the sign surface, ensuring image quality and improving the clarity of character edges.

[0061] The image acquisition and control unit monitors the wheel pair position in real time through photoelectric sensors. When the wheel enters the preset detection area, a synchronization control signal is triggered to drive the image acquisition units on both sides to capture images of the wheel pair identification plates on the left and right sides.

[0062] The data processing and analysis unit is responsible for image post-processing and information extraction. Its processing flow includes: Image preprocessing: Denoising and contrast enhancement are performed on the original image to improve image quality; Signage region extraction: Based on color, texture, or shape features, image segmentation algorithms are used to locate the area where the sign is located; Text recognition: Using deep learning-based optical character recognition (OCR) algorithms, the text areas and their orientations on signs are located for character recognition; Information comparison and management: The identification results are matched with the wheelset information pre-stored in the background database to automatically determine key parameters such as the wheelset's operating cycle and cumulative mileage, and update maintenance records to achieve intelligent tracking and early warning of wheelset status.

[0063] Compared to existing technologies, the system provided in this embodiment: adopts a dual-sided symmetrical image acquisition layout, which can simultaneously identify and record wheel identification information on both the left and right sides; combines shadowless ring illumination and image enhancement technology to significantly improve imaging consistency in complex environments; achieves automatic detection and correction recognition of text arranged in any direction, eliminating the need for manual adjustment of wheel pair positions and straightening of identification plates, thus enhancing the system's adaptability to changes in nameplate posture; and improves wheelset management efficiency and data accuracy through automated information acquisition and database linkage, while reducing the cost of manual intervention.

[0064] Example 4: In some possible scenarios, a portable mobile identification system can be realized based on the aforementioned omnidirectional text information recognition system for wheelset signage.

[0065] The portable mobile identification system is applicable to scenarios including: emergency repairs in the field, small maintenance stations, and emergency wheel inventory (without fixed inspection stations).

[0066] In addition, the following changes can be made to the structure of this portable mobile identification system: remove the fixed installation structure and integrate the image acquisition unit (small industrial camera + portable shadowless light source) and data processing and analysis unit (lightweight edge computing terminal) into the handheld gun-shaped shell; add a display screen and a trigger button, and simplify the image acquisition control unit to a built-in distance sensor; workflow: the operator holds the device and aligns it with the wheel pair identification sign. After the distance sensor detects the effective distance (0.3-1m), the operator presses the trigger button, and the device simultaneously takes a picture and completes the identification. The result is displayed on the screen in real time and can be uploaded to the mobile terminal via Bluetooth.

[0067] Example 5: Based on the above embodiments, this application provides a specific application example for the method of omnidirectional text information recognition and health status monitoring for wheelset identification signs.

[0068] Step 1: Wheelset position monitoring and triggering: When the wheelset enters the preset detection area along the track, the photoelectric sensor detects the wheelset signal and immediately transmits it to the synchronization controller of the image acquisition control unit. The synchronization controller quickly generates a synchronization trigger command to ensure that the image acquisition units on both sides start working at the same time, avoiding information loss due to shooting time difference.

[0069] Step 2: Bilateral synchronous image acquisition: After receiving the command, the image acquisition units on both sides automatically adjust the brightness of the shadowless ring light source according to the ambient light intensity (e.g., 500-800 lux in strong light environment, 1500-2000 lux in weak light environment) to suppress reflections and shadows on the surface of the sign; the high-resolution industrial camera simultaneously captures images of the signs on both sides of the wheelset, generating two clear images per group of shots (corresponding to the signs on both sides respectively), and transmits them to the data processing and analysis unit in real time through the high-speed communication interface.

[0070] Step 3: Image preprocessing and signage region extraction: After receiving the original image, the data processing and analysis unit first performs preprocessing: removing image noise through Gaussian filtering algorithm, enhancing local contrast through histogram equalization technology, and eliminating the effects of uneven lighting; then, based on color and shape features, roughly locating the sign area through image segmentation algorithm, and then combining edge detection and morphological operations to accurately locate the smallest bounding rectangle of the sign and remove background interference.

[0071] Step 4: Text detection, correction, and OCR recognition: A deep learning-based text detection algorithm is used to scan the text blocks within the sign area and calculate the main direction (tilt angle) of the text. The text area is rotated and corrected according to the tilt angle, turning text in any direction into a horizontal direction. The deep learning OCR module is then activated to recognize characters in the corrected text area and output the recognition result string and confidence score (the confidence score threshold is set to 90%, and if it is lower than the threshold, the detection is repeated).

[0072] Step 5: Information Comparison and Management The data processing and analysis unit matches the identification results with the wheelset archive information pre-stored in the background database, automatically calculates the cumulative running mileage and usage cycle of the wheelset, and determines whether it has reached the maintenance or scrapping threshold; it uploads the identification results, matching conclusions, detection time and other information to the background database, updates the wheelset maintenance records, supports historical traceability and report generation; if the matching fails or the confidence level is insufficient, it sends a reminder to the operation and maintenance terminal and waits for manual review.

[0073] Example 6: Another embodiment of this application relates to an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the methods described in the above embodiments.

[0074] The memory and processor are connected via a bus, which can include any number of interconnecting buses and bridges, connecting various circuits of one or more processors and memories. The bus can also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and will not be described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor is transmitted over the wireless medium via an antenna, which further receives data and transmits it to the processor.

[0075] The processor manages the bus and handles general processing, and also provides various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory, on the other hand, is used to store data used by the processor during operation.

[0076] Example 7: Another embodiment of this application relates to a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the method embodiments described above.

[0077] That is, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0078] In some embodiments of this application, a computer program product is also provided, including a computer program that, when executed by a processor, implements the steps of the methods described in the above embodiments.

[0079] Those skilled in the art will understand that the above embodiments are specific embodiments for implementing this application, and in practical applications, various changes can be made to them in form and detail without departing from the spirit and scope of this application.

Claims

1. A method for omnidirectional text information recognition of wheelset identification signs, characterized in that, include: In response to the detection that the train wheelset has reached the preset detection area, a synchronous data acquisition command is issued; In response to the synchronous acquisition command, the lighting environment of the signs on both sides of the train wheelset is adjusted synchronously, and an image acquisition operation is performed to obtain two signs images; For each sign image, the corresponding text region image is extracted based on the text region, and optical character recognition is performed on the text region image to obtain the corresponding recognition result.

2. The method according to claim 1, characterized in that, The method further includes: If the recognition result does not meet the preset conditions, a synchronization acquisition command is reissued to re-acquire the sign image and perform optical character recognition, thereby obtaining the recognition result again.

3. The method according to claim 1, characterized in that, The steps for adjusting the lighting environment include: Obtain the current ambient light intensity of the environment where the sign is located; Based on the correspondence table between the current ambient light intensity and the preset supplementary light amount, the target supplementary light amount is determined; Based on the target supplementary light amount, a supplementary light control command is generated to drive the supplementary light device to perform adaptive supplementary light on the train wheelsets.

4. The method according to claim 3, characterized in that, The acquisition of the current ambient light intensity of the environment where the sign is located includes: The current ambient light intensity of the environment where the sign is located is obtained at preset time intervals.

5. The method according to claim 1, characterized in that, The steps for determining the recognition result include: The text region is obtained from the sign image using a deep learning-based text detection algorithm, and the tilt angle of the text region is calculated. The text area is rotated and corrected according to the tilt angle so that the text in the text area is in a horizontal direction; The corrected text region is subjected to character recognition using OCR to obtain the recognition result.

6. The method according to claim 1, characterized in that, The recognition result includes the text recognition result and the corresponding confidence score; the method further includes: If the confidence level of each of the recognition results is not less than the confidence level threshold, retrieval information is generated based on the text recognition result corresponding to the manufacturing information sign; wherein, the sign includes manufacturing information signs and maintenance information signs; Based on the retrieved information and the preset wheelset file information, the cumulative operating mileage and usage cycle of the train wheelset are determined; Based on the cumulative operating mileage and the usage cycle, it is determined whether the train wheelset has reached the maintenance or scrapping threshold, and if the maintenance or scrapping threshold is reached, a health status reminder message is issued.

7. The method according to claim 6, characterized in that, The method further includes: The retrieval results of the search information, the identification results corresponding to each sign, and the current detection time are stored in the database to update the maintenance records of the train wheelsets.

8. A 360-degree text information recognition system for wheelset identification signs, characterized in that, include: The image acquisition and control unit is used to issue a synchronous acquisition command in response to the detection that the train wheelset has reached the preset detection area; An image acquisition unit, communicatively connected to the image acquisition unit, is used to respond to the synchronous acquisition command to simultaneously perform image acquisition operations on both sides of the train wheelset for the signboard, thereby obtaining two signboard images; A data processing and analysis unit, which is communicatively connected to the data processing and analysis unit, is used to extract the corresponding text region image based on the text region of each sign image, and to perform optical character recognition on the text region image to obtain the corresponding recognition result.

9. The system according to claim 8, characterized in that, The recognition result includes text recognition result and corresponding confidence score; the system further includes: a data retrieval processing unit, a data retrieval unit, and a health status determination unit; wherein: The retrieval data processing unit is used to generate retrieval information based on the text recognition results of each recognition result, provided that the confidence level of each recognition result is not less than the confidence level threshold. The data retrieval unit is communicatively connected to the retrieval data processing unit and is used to determine the cumulative operating mileage and usage cycle of the train wheelset based on the retrieval information and the preset wheelset file information. The health status determination unit is communicatively connected to the data retrieval unit and is used to determine whether the train wheelset has reached the maintenance or scrapping threshold based on the cumulative operating mileage and the usage cycle, and to issue a health status reminder message if the maintenance or scrapping threshold is reached.

10. The system according to claim 9, characterized in that, The system also includes a light sensor, a supplementary lighting control unit, and a supplementary lighting device; wherein: The light sensor is used to obtain the current ambient light intensity of the environment in which the sign is located; The data processing and analysis unit is communicatively connected to the light sensor and is also used to determine the target supplementary light amount based on the current ambient light intensity. The supplementary lighting control unit is communicatively connected to the data processing unit and is used to generate supplementary lighting control commands based on the target supplementary lighting amount to drive the supplementary lighting device to adaptively supplement the signboard. The supplementary lighting device is communicatively connected to the supplementary lighting control unit and is used to emit light to the train wheelsets to meet preset imaging requirements.

Images