Tire character recognition method, device and electronic equipment

Abstract

The present application discloses a tire character recognition method, device, and electronic device, relating to the field of image processing technology. The method comprises: capturing images of multiple areas of a target tire to obtain multiple images, and splicing the multiple images to obtain an annular image; converting the annular image into a horizontal image through polar coordinate conversion; identifying character information on the horizontal image, and determining the category information of the currently recognized character based on the character information; comparing the category of the character with the standard category and outputting character position difference information and character content difference information. Based on the above method, when recognizing tire characters, using a 3D camera to obtain an image of the tire can improve the accuracy of character recognition, and can realize the recognition of ordinary text, images, and special characters.

Description

Technical Field

[0001] The present application mainly relates to the field of image processing technology, and in particular to a tire character recognition method, device and electronic equipment. Background Art

[0002] With the rapid development of intelligent manufacturing, more and more advanced manufacturing technologies are being applied to production. Among them, Optical Character Recognition (OCR) has become an important technical means in industrial production. It can convert characters printed on industrial products into text format for subsequent analysis and processing.

[0003] Tire character recognition is a key application for OCR technology. Character markings on tire surfaces often reflect the performance and characteristics of rubber tires on the vulcanization and molding lines. Currently, tire character detection is mostly performed manually, but this is difficult to achieve due to high-volume production and complex working environments.

[0004] When using OCR technology to recognize tire characters, the most common mode is to recognize characters on the tire mold rather than on the tire itself. This is because tire molds have advantages such as a simple background and regular structure, making them effective for recognizing large and medium-sized characters. However, when recognizing full characters, small character recognition is less effective due to low imaging accuracy and poor grayscale. Furthermore, when detecting character defects, the characters on the tire are typically located first, followed by character accuracy testing. Existing technologies generally use template matching to compare the image to be tested with a template image to determine the degree of character matching. However, due to the non-fixed nature of tire printed characters, this method is less versatile, resulting in poor accuracy when detecting character defects. Summary of the Invention

[0005] The present application provides a tire character recognition method, device and electronic device, which realizes text positioning recognition, pattern recognition and special character recognition by adopting a 3D imaging method to capture tire images, thereby solving the problem of poor character recognition effect when performing tire character recognition.

[0006] In a first aspect, the present application provides a tire character recognition method, the method comprising:

[0007] Capturing images of multiple areas of the target tire to obtain multiple images, and stitching the multiple images to obtain an annular image;

[0008] Converting the annular image into a horizontal image by polar coordinate conversion;

[0009] Recognize character information on the horizontal image, and determine category information of the currently recognized character based on the character information, wherein the category information includes character position information and character content information of the character in the horizontal image;

[0010] The category of the character is compared with a standard category and character position difference information and character content difference information are output.

[0011] By using the above method to capture the image of the target tire and further converting the annular image into a horizontal image, the characters can be recognized and the content and position differences between the current characters and the standard category characters can be determined.

[0012] In an optional implementation, the identifying character information on the horizontal image includes:

[0013] Locating the current row in the horizontal image;

[0014] Dividing the text line into multiple contents to be detected;

[0015] Detecting the multiple contents to be detected, and combining the detection results of the multiple contents to be detected into one content to be detected;

[0016] Inputting the content to be detected into a recognition network for recognition to obtain a character recognition result in the text line;

[0017] Using the character recognition result as the character information;

[0018] Through the above method, the text line is recognized in a sliding block manner, and the detection network is used to detect multiple contents to be detected in parallel, thereby improving the accuracy of character detection.

[0019] In an optional implementation, the identifying character information on the horizontal image includes:

[0020] Use deep learning retrieval network to conduct online training for special characters and establish feature templates in the database;

[0021] Matching the special character to be queried in the horizontal image with the feature template to obtain a detection result of the special character to be queried;

[0022] The detection result of the special character to be queried is used as the character information.

[0023] Through the above method, special characters that have not appeared before can be updated into the database by adopting the online training method, thereby ensuring the accuracy in recognizing special characters.

[0024] In an optional embodiment, after comparing the category of the character with the standard category and outputting the character position difference information and the character content difference information, the method further includes:

[0025] Inputting the query pattern and the pattern template in the horizontal image into a shared feature extractor to extract query pattern features of the query pattern and pattern template features of the pattern template respectively;

[0026] The query pattern feature is compared with the pattern template feature, and if the similarity between the query pattern feature and the pattern template feature is greater than a preset threshold, the position difference information of the query pattern is output.

[0027] In summary, the use of a 3D camera to capture images of the target tire improves the image accuracy, images the tire characters on the three-dimensional image, and maps the height difference into grayscale values. The higher the camera accuracy, the better the character imaging quality. This scheme realizes the recognition of ordinary characters, patterns, and special characters on the tire. During image recognition, the function of online updating of new category pattern samples can also be realized through the meta-feature convolution module and the meta-feature update module.

[0028] In a second aspect, the present application provides a tire character recognition device, the device comprising:

[0029] a processing module, configured to capture images of multiple areas of the target tire to obtain multiple images, and to splice the multiple images to obtain an annular image;

[0030] A conversion module, configured to convert the annular image into a horizontal image by polar coordinate conversion;

[0031] a recognition module, configured to recognize character information on the horizontal image and determine a category of a currently recognized character based on the character information;

[0032] The output module is used to compare the category of the character with the standard category and output character position difference information and character content difference information.

[0033] In an optional implementation, the identification module is further configured to:

[0034] locating a line of text in the horizontal image;

[0035] Dividing the text line into multiple contents to be detected;

[0036] Detecting the multiple contents to be detected, and combining the detection results of the multiple contents to be detected into one content to be detected;

[0037] Inputting the content to be detected into a recognition network for recognition to obtain a character recognition result in the text line;

[0038] The character recognition result is used as the character information.

[0039] In an optional implementation, the identification module is further configured to:

[0040] Use deep learning retrieval network to conduct online training for special characters and establish feature templates in the database;

[0041] Matching the special character to be queried in the horizontal image with the feature template to obtain a detection result of the special character to be queried;

[0042] The detection result of the special character to be queried is used as the character information.

[0043] In an optional implementation, the identification module is further configured to:

[0044] Inputting the query pattern and the pattern template in the horizontal image into a shared feature extractor to extract query pattern features of the query pattern and pattern template features of the pattern template respectively;

[0045] The query pattern feature is compared with the pattern template feature. If the similarity between the query pattern feature and the pattern template feature is greater than a preset threshold, the position difference information of the query pattern is output.

[0046] In a third aspect, the present application provides an electronic device, comprising:

[0047] Memory for storing computer programs;

[0048] The processor is configured to implement the steps of the above-mentioned tire character recognition method when executing the computer program stored in the memory.

[0049] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned tire character recognition method are implemented.

[0050] For each of the above-mentioned aspects from the second to the fourth aspects and the technical effects that may be achieved by each of the aspects, please refer to the above-mentioned description of the technical effects that can be achieved by the first aspect and the various possible solutions in the first aspect, and no further details will be given here. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A flowchart of a tire character recognition method provided in an embodiment of the present application;

[0052] Figure 2 A schematic diagram of multi-region image acquisition provided in an embodiment of the present application;

[0053] Figure 3 A schematic diagram of a coordinate system for polar coordinate conversion provided in an embodiment of the present application;

[0054] Figure 4 A schematic diagram of identifying a small-size area of ​​a text line provided in an embodiment of the present application;

[0055] Figure 5 A schematic diagram of a tire character recognition device provided in an embodiment of the present application;

[0056] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail with reference to the accompanying drawings. The specific operating methods in the method embodiments can also be applied to device embodiments or system embodiments. It should be noted that in the description of the present application, "multiple" is understood as "at least two". "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the following three situations: A exists alone, A and B exist at the same time, and B exists alone. A is connected to B, which can represent the following two situations: A is directly connected to B and A is connected to B through C. In addition, in the description of the present application, words such as "first" and "second" are only used to distinguish the purpose of description, and cannot be understood as indicating or implying relative importance, nor can they be understood as indicating or implying order.

[0058] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0059] In industrial production, optical character recognition (OCR) technology is a crucial tool for converting characters printed on industrial products into text format for subsequent analysis and processing. OCR technology uses industrial cameras to capture real-time images of products on the production line. Then, using visual processing software, it processes the images, identifies characters within them, and extracts information from the images. This technology plays a crucial role in product quality inspection on the production line, significantly improving product quality and avoiding oversights that can occur with manual inspections.

[0060] Tire character recognition is an important application scenario for OCR technology. During the first-article inspection, defects in the characters on the tire can be detected in a timely manner to avoid the same character defects on subsequent products. Currently, the use of OCR technology to achieve tire character recognition mainly involves character recognition of the tire mold rather than character recognition of the tire itself. When recognizing characters on the tire mold, the recognition effect of large and medium-sized characters is better. When achieving full character recognition, the recognition effect of small characters is poor due to low imaging accuracy of small characters and poor grayscale. In addition, when performing character defect detection, the tire format printing characters are not fixed and have poor versatility, resulting in poor accuracy when performing character defect detection.

[0061] To address the aforementioned issues, embodiments of the present application provide a tire character recognition method, device, and electronic device. To address issues such as slow character imaging and poor grayscale during tire character recognition, a 3D camera is used to image the target tire's characters on a three-dimensional image and map the height differences into grayscale values. As camera accuracy improves, the character imaging quality is significantly enhanced, enabling recognition of conventional characters, patterns, and special characters.

[0062] The technical solution of this application is further described below through specific embodiments. Figure 1 As shown in FIG, a tire character recognition method provided by this application, the implementation process of the method is as follows:

[0063] S1, collecting images of multiple areas of the target tire to obtain multiple images, and stitching the multiple images to obtain a ring image;

[0064] First, images of multiple areas of the target tire are captured. Specifically, a 3D camera is used to capture images of the target tire. A 3D camera can capture three-dimensional information about the target object, enabling functions that 2D vision cannot, such as inspecting the product's height, flatness, and volume. Due to the limited field of view of a 3D camera, a single 3D camera cannot capture a complete image of the target tire. Therefore, different numbers of 3D cameras are required for different tire sizes.

[0065] For example, when capturing images of truck tires, which may be 20 inches in diameter, more 3D cameras are needed to capture images of larger diameter tires. In this case, four 3D cameras might be needed to complete the tire image capture. However, when capturing images of car tires, which are smaller than truck tires, only three 3D cameras might be needed.

[0066] It should be noted that the number of 3D cameras is only an example. In actual implementation, the number of 3D cameras needs to be adjusted according to the specific size of the tire. Figure 2 As shown, the target tire is divided into multiple areas, each area corresponds to a 3D camera, and when performing image acquisition, each 3D camera acquires an image of the corresponding area.

[0067] For example, 3D camera 1 corresponds to area 1, 3D camera 2 corresponds to area 2, and 3D camera 3 corresponds to area 3. After the 3D cameras collect images of their respective corresponding areas, multiple images are obtained.

[0068] Furthermore, after capturing multiple images of various regions of the target tire, the images captured by the corresponding cameras for each region are stitched together to create a complete annular image. When stitching together these multiple images, it is first necessary to remove outliers. When using a 3D camera to capture images of entities in reverse engineering, the measurement process is affected by human factors and the environment, which can introduce outliers—small pieces of point clouds and discrete points that are far from the main point cloud. The presence of outliers can affect the quality of subsequent modeling, so they must be removed after stitching the multiple images. After removing the outliers, the image is projected into a 2D image to create an annular image.

[0069] The above method divides the target tire into multiple regions when capturing images. Using a 3D camera to capture images of each corresponding region ensures that the image captured in each region has relatively high accuracy. This avoids the problems of poor imaging quality and low resolution that can occur when using a single 3D camera to capture images of the target tire.

[0070] S2, converts the annular image into a horizontal image through polar coordinate transformation;

[0071] like Figure 3 As shown, all points in polar coordinates can be uniquely determined by the parameters θ and ρ. For a circle, ρ is the fixed radius. If converted to a rectangular coordinate system, it is a straight line parallel to the horizontal axis. When the horizontal coordinate is transformed, the vertical coordinate remains unchanged.

[0072] Furthermore, when converting the annular image into a horizontal image, the circumscribed circle radius, inscribed circle radius and center of the annular image are first determined, then the annular area of ​​the annular image is determined based on the circumscribed circle radius, inscribed circle radius and center of the circle, and threshold processing is performed on the annular image, and finally the display area of ​​the horizontal image is obtained through polar coordinate conversion.

[0073] Through the above method, the annular image can be converted into a horizontal image using polar coordinate transformation.

[0074] S3, identifying character information on the horizontal image, and determining category information of the currently recognized character based on the character information;

[0075] In an optional embodiment, when recognizing a text line of regular characters, a deep learning text localization network is used to locate the text line in a horizontal image. Furthermore, the text line is divided into multiple sections to be detected, and each of the sections to be detected is detected separately. After obtaining a detection result for each section to be detected, the detection results of the multiple sections to be detected are merged into a single section to be detected. This single section to be detected is then input into a recognition network for recognition, thereby obtaining a character recognition result for the text line.

[0076] Specifically, when recognizing text lines in horizontal images, the horizontal image is obtained by converting the circular image through polar coordinates. Therefore, the horizontal image has a high resolution and cannot fully detect text lines in the entire horizontal image. In practice, a sliding partitioning method is used to divide the horizontal image into multiple sections to be detected. A detection network is then used to perform parallel detection on these sections, allowing for simultaneous detection and obtaining detection results.

[0077] In the embodiments of this application, when detecting tire characters, the actual physical height of the characters on the tire is small. For example, the actual size of letters or Chinese characters on a car tire is only a few centimeters. Therefore, after the image is captured by a 3D camera and projected into a 2D image, the accuracy of some characters on the image will be relatively low. In addition, the tire has a large amount of shading, which also interferes with character detection. In the specific implementation, to avoid the above problems, a regression-based detection network is used for character detection.

[0078] For example, Figure 4 As shown in the figure, for example, if the content of a text line is "campus", a small text box is constructed to improve the accuracy of character detection. The character "campus" is divided into multiple detection areas, such as area 1, area 2, area 3, and so on to area N. The number of areas N is determined by the length of the text line and the length of each detection area.

[0079] During recognition, character detection is performed simultaneously on the divided areas 1, 2, 3, and N to obtain the detection results of each character area, and then the results of multiple detection areas are merged into a complete character, and the complete characters are further merged into a content to be detected, that is, a complete text line.

[0080] like Figure 4For example, the text line shown in the figure is divided into multiple sections to be detected, and each section is recognized simultaneously. For example, the sections to be detected in regions 1, 2, 3, and 4 are recognized and then merged to obtain the letter C. Similarly, the sections to be detected in the other regions are recognized to obtain the remaining letters, and after all letters are recognized, they are merged to obtain the complete text line content.

[0081] In the embodiment of the present application, before character detection, a character dictionary model is established for each tire type based on the production information provided by the factory. Specifically, each tire type has different characters on its surface. For example, information such as tire size and load varies, and the numbers marked on the tire surface also vary. Therefore, a character dictionary model needs to be established based on the character information of each tire type.

[0082] The resulting content to be detected is then fed into a Lexicon-based convolutional recurrent neural network (CRNN). The CRNN then identifies the text detection box and produces end-to-end text line recognition results. For example, if the text line contains the word "campus," the CRNN can identify the entire line.

[0083] Furthermore, the category information of the currently recognized character is determined based on the character information on the horizontal image. Specifically, there are character categories such as DOT, load, warning, etc. on the tire.

[0084] For example, the DOT information on a certain tire model is DOT 4B9Z 747R 1822. The first eight digits indicate the tire's origin and factory code, while the last four digits indicate the tire's production date. Another example is a tire with a load index of 91, indicating a maximum load capacity of 615 kg. Specifically, the location of the category information corresponding to each tire specification is fixed on the tire. Therefore, the tire's category information can be used to determine the location and content of the characters on the tire.

[0085] Through the above method, a regression-based detection network is used to construct a small-sized text box to recognize characters. Finally, the recognized contents are merged to achieve end-to-end recognition of the entire line of text, ensuring the accuracy of tire character recognition.

[0086] In the embodiment of the present application, there are usually some special characters when performing tire character recognition, such as the tire's product name, brand name, etc. Most of the special characters on the tire are designed cursive characters, so it is difficult to perform character recognition.

[0087] In an optional embodiment, a deep learning retrieval network is used to perform online training on special characters, and a feature template is established in the database.

[0088] Specifically, when performing character recognition, for patterns or logos that have never appeared, special characters are trained online through a deep learning retrieval network. Through online training, special characters that have never appeared are extracted and stored in the database.

[0089] Furthermore, the deep learning retrieval module compares the special characters to be queried in the horizontal image with the feature templates in the database to obtain the retrieval results of the special characters in the horizontal image. Finally, the detection results of the special characters to be queried are used as the character information currently retrieved in the horizontal image.

[0090] The above method can be used to establish a feature template for special characters by performing online training on special characters. Furthermore, when recognizing special characters, a convolutional network is used to extract the features of the query image and the features of the feature template to generate a feature map. The location information and category of the special characters are then determined based on the feature map.

[0091] In an optional embodiment, the query pattern and the pattern template in the horizontal image are input into a shared feature extractor to extract query pattern features of the query pattern and pattern template features of the pattern template respectively.

[0092] For example, there are a large number of patterns on tires, such as CCC standards and brand logos. These patterns are complex and changeable, making it difficult to obtain samples using online training methods.

[0093] Therefore, when performing pattern recognition, it is mainly based on the deep learning network YoloV5, and the recognition of the pattern is achieved through the meta-feature convolution module in the shared feature extractor. Specifically, when identifying a certain type of pattern on the tire, the features of the pattern to be queried and the pattern template are first extracted to obtain a feature map. Specifically, the category and position information of the pattern to be queried and the pattern template are determined. Further, the features of the pattern to be queried are compared with the features of the pattern template. If the similarity between the features of the pattern to be queried and the features of the pattern template is greater than a preset threshold, for example, the preset threshold is set to 95%. When the similarity between the features of the pattern to be queried and the features of the pattern template is greater than 95%, the position difference information of the pattern to be queried can be output.

[0094] By comparing the features, the category and position information of the pattern to be queried are determined, and after the recognition is completed, the position difference information of the pattern to be queried can be output.

[0095] In the embodiment of the present application, the meta-feature updating module can realize the functions of small sample patterns and online updating of new category samples.

[0096] Specifically, when performing pattern recognition, if a new category of pattern template needs to be added, the features of the new category of pattern template can be extracted online through a shared feature extractor, and the newly extracted features can be added to the pattern template through a meta-feature update module.

[0097] The above method can realize the recognition of small sample patterns, and can also realize the function of online updating of new category patterns through the meta-feature updating module.

[0098] S4, comparing the character category with the standard category and outputting character position difference information and character content difference information.

[0099] After the character recognition is completed, the character category is determined, and the recognized character category is further compared with the standard category.

[0100] For example, if the recognized character category is determined to be DOT, the recognized DOT content is compared with the standard DOT content on the tire to determine the difference from the standard content. Based on the comparison result, the character position information and content difference information between the recognition result and the standard category are output. The character difference information is the character difference obtained after comparing the recognition result with the standard category. For example, the recognized DOT content is DOT 4B9Z 747R 1822, while the standard category DOT content is DOT 4B9Z 747R1622. Therefore, the recognized character content is misprinted. For another example, the standard category DOT content should be in area A on the tire, but the recognized DOT content deviates from area A. Therefore, the character position difference information in the horizontal image can be output.

[0101] Through the above method, the category of the character is compared with the standard category to obtain the difference between the current character and the standard category character, and the character position difference information and the character content difference information are obtained, so that it can be judged whether the currently recognized character has problems such as missing printing, misprinting or printing position error based on the character position difference information and the character content difference information.

[0102] Based on the same inventive concept, an embodiment of the present application also provides a tire character recognition device for recognizing tire characters, which can improve the accuracy of tire character recognition and realize the recognition of ordinary text, patterns and special characters.

[0103] Reference Figure 5 As shown, the device includes:

[0104] The processing module 501 is configured to capture images of multiple regions of the target tire to obtain multiple images, and to stitch the multiple images to obtain an annular image;

[0105] A conversion module 502, configured to convert the annular image into a horizontal image by polar coordinate conversion;

[0106] a recognition module 503 for recognizing character information on the horizontal image and determining category information of a currently recognized character based on the character information;

[0107] The output module 504 is configured to compare the character category with a standard category and output character position difference information and character content difference information.

[0108] In an optional implementation, the identification module is further configured to:

[0109] locating a line of text in the horizontal image;

[0110] Dividing the text line into multiple contents to be detected;

[0111] Detecting the multiple contents to be detected, and combining the detection results of the multiple contents to be detected into one content to be detected;

[0112] Inputting the content to be detected into a recognition network for recognition to obtain a character recognition result in the text line;

[0113] The character recognition result is used as the character information.

[0114] In an optional implementation, the identification module is further configured to:

[0115] Use deep learning retrieval network to conduct online training for special characters and establish feature templates in the database;

[0116] Matching the special character to be queried in the horizontal image with the feature template to obtain a detection result of the special character to be queried;

[0117] The detection result of the special character to be queried is used as the character information.

[0118] In an optional implementation, the identification module is further configured to:

[0119] Inputting the query pattern and the pattern template in the horizontal image into a shared feature extractor to extract query pattern features of the query pattern and pattern template features of the pattern template respectively;

[0120] The query pattern feature is compared with the pattern template feature. If the similarity between the query pattern feature and the pattern template feature is greater than a preset threshold, the position difference information of the query pattern is output.

[0121] It should be noted that the above-mentioned device provided in the embodiment of the present application can implement all the method steps in the above-mentioned tire character recognition method embodiment and can achieve the same technical effects. The parts and beneficial effects of this embodiment that are the same as those of the method embodiment will not be described in detail here.

[0122] Based on the same inventive concept, an electronic device is also provided in the embodiment of the present application. The electronic device can realize the function of the aforementioned tire character recognition method. Figure 6 As shown, the electronic device includes:

[0123] At least one processor 601, and a memory 602 connected to the at least one processor 601. The specific connection medium between the processor 601 and the memory 602 is not limited in the embodiment of the present application. Figure 6 In the example, the processor 601 and the memory 602 are connected via a bus 600. Figure 6 The bus 600 can be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, Figure 6 The diagram is represented by only one thick line, but this does not mean that there is only one bus or one type of bus. Alternatively, the processor 601 may also be referred to as a controller, without limitation to the name.

[0124] In the embodiment of the present application, the memory 602 stores instructions that can be executed by at least one processor 601. The at least one processor 601 can execute the tire character recognition method discussed above by executing the instructions stored in the memory 602. The processor 601 can implement Figure 5 The functions of each module in the device shown.

[0125] Among them, the processor 601 is the control center of the device, which can use various interfaces and lines to connect the various parts of the entire control device, and monitor the device as a whole by running or executing instructions stored in the memory 602 and calling data stored in the memory 602, the various functions of the device and processing data.

[0126] In one possible design, processor 601 may include one or more processing units. Processor 601 may integrate an application processor and a modem processor. The application processor primarily processes the operating system, user interface, and application programs, while the modem processor primarily processes wireless communications. It is understood that the modem processor may not be integrated into processor 601. In some embodiments, processor 601 and memory 602 may be implemented on the same chip. In some embodiments, they may also be implemented on separate chips.

[0127] Processor 601 can be a general-purpose processor, such as a central processing unit (CPU), a digital signal processor, an application-specific integrated circuit, a field-programmable gate array or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component, and can implement or execute the various methods, steps, and logic diagrams disclosed in the embodiments of this application. A general-purpose processor can be a microprocessor or any conventional processor. The steps of the tire character recognition method disclosed in the embodiments of this application can be directly implemented and executed by a hardware processor, or by a combination of hardware and software modules within the processor.

[0128] The memory 602 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer executable programs and modules. The memory 602 may include at least one type of storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory, a random access memory (Random Access Memory, RAM), a static random access memory (Static Random Access Memory, SRAM), a programmable read-only memory (Programmable Read Only Memory, PROM), a read-only memory (Read Only Memory, ROM), an electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, EEPROM), a magnetic memory, a disk, an optical disk, etc. The memory 602 is any other medium that can be used to carry or store a desired program code in the form of an instruction or data structure and can be accessed by a computer, but is not limited thereto. The memory 602 in the embodiment of the present application can also be a circuit or any other device that can realize a storage function, for storing program instructions and / or data.

[0129] By designing and programming the processor 601, the code corresponding to the tire character recognition method described in the above embodiment can be fixed into the chip, so that the chip can execute the code when running. Figure 1The steps of the tire character recognition method of the embodiment shown are as follows: How to design and program the processor 601 is well known to those skilled in the art and will not be described in detail here.

[0130] Based on the same inventive concept, an embodiment of the present application further provides a storage medium storing computer instructions. When the computer instructions are executed on a computer, the computer executes the tire character recognition method discussed above.

[0131] In some possible implementations, various aspects of the tire character recognition method provided in the present application may also be implemented in the form of a program product, which includes program code. When the program product is run on a device, the program code is used to cause the control device to execute the steps of the tire character recognition method according to various exemplary embodiments of the present application described above in this specification.

[0132] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0133] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0134] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0135] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0136] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.

Claims

1. A tire character recognition method, characterized in that: For realizing the recognition of regular characters, patterns and special characters, the method includes: Capturing multiple regions of a target tire to obtain multiple images, and stitching the multiple images to obtain a ring-shaped image; wherein the multiple images are acquired by a 3D camera and include three-dimensional information of the target tire; Converting the annular image into a horizontal image by polar coordinate conversion; Recognizing character information on the horizontal image, and determining category information of a currently recognized character based on the character information, wherein the category information includes character position information and character content information of the character in the horizontal image, and the character includes the regular character and the special character; Comparing the character category with a standard category and outputting character position difference information and character content difference information; Inputting the query pattern and the pattern template in the horizontal image into a shared feature extractor to extract query pattern features of the query pattern and pattern template features of the pattern template respectively; The query pattern feature is compared with the pattern template feature. If the similarity between the query pattern feature and the pattern template feature is greater than a preset threshold, the position difference information of the query pattern is output.

2. The method according to claim 1, wherein The identifying of character information on the horizontal image includes: locating a line of text in the horizontal image; Dividing the text line into multiple contents to be detected; Detecting the multiple contents to be detected, and combining the detection results of the multiple contents to be detected into one content to be detected; Inputting the content to be detected into a recognition network for recognition to obtain a character recognition result in the text line; The character recognition result is used as the character information.

3. The method according to claim 1, wherein The identifying of character information on the horizontal image includes: Use deep learning retrieval network to conduct online training for special characters and establish feature templates in the database; Matching the special character to be queried in the horizontal image with the feature template to obtain a detection result of the special character to be queried; The detection result of the special character to be queried is used as the character information.

4. A tire character recognition device, characterized in that: For realizing the recognition of conventional characters, patterns and special characters, the device comprises: a processing module, configured to acquire images of multiple regions of a target tire to obtain multiple images, and to stitch the multiple images together to obtain a ring-shaped image; wherein the multiple images are acquired by a 3D camera and include three-dimensional information of the target tire; A conversion module, configured to convert the annular image into a horizontal image by polar coordinate conversion; a recognition module, configured to recognize character information on the horizontal image, and determine category information of a currently recognized character based on the character information, wherein the category information includes character position information and character content information of the character in the horizontal image, and the character includes the regular character and the special character; an output module, configured to compare the category of the character with a standard category and output character position difference information and character content difference information; The recognition module is further configured to input the pattern to be queried and the pattern template in the horizontal image into the shared feature extractor, and respectively extract the query pattern features of the pattern to be queried and the pattern template features of the pattern template; The query pattern feature is compared with the pattern template feature. If the similarity between the query pattern feature and the pattern template feature is greater than a preset threshold, the position difference information of the query pattern is output.

5. The device according to claim 4, characterized in that The identification module is further configured to: locating a line of text in the horizontal image; Dividing the text line into multiple contents to be detected; Detecting the multiple contents to be detected, and combining the detection results of the multiple contents to be detected into one content to be detected; Inputting the content to be detected into a recognition network for recognition to obtain a character recognition result in the text line; The character recognition result is used as the character information.

6. The device according to claim 4, characterized in that The identification module is further configured to: Use deep learning retrieval network to conduct online training for special characters and establish feature templates in the database; Matching the special character to be queried in the horizontal image with the feature template to obtain a detection result of the special character to be queried; The detection result of the special character to be queried is used as the character information.

7. An electronic device, characterized in that: include: Memory for storing computer programs; A processor, configured to implement the method steps of any one of claims 1 to 3 when executing the computer program stored in the memory.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method steps according to any one of claims 1 to 3 are implemented.

Images