Road image generation method and device, equipment and storage medium

Abstract

The invention discloses a road image generation method and device, equipment and a storage medium. The method comprises the steps of obtaining an urban road image of a to-be-drawn map; converting the urban road image into a binary image; inputting the binary image into the trained generation agent to obtain predictive coding information of the binary image; inputting the predictive coding information into the trained discriminant agent to obtain a connectivity evaluation result about the predictive coding information; when a connectivity evaluation result reaches a preset requirement, carrying out inverse coding processing on the prediction coding information to obtain a prediction road network diagram; and when the connectivity evaluation result does not meet the preset requirement, triggering the generation intelligent agent to generate predictive coding information of the newly generated binary image. The road image generation efficiency and accuracy can be improved.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to a method, device, equipment and storage medium for generating a road image. Background Art

[0002] With the acceleration of urbanization, the urban population and the number of vehicles are increasing, urban transportation is facing increasingly severe challenges, and the demand for intelligent transportation systems is also growing. The generation and optimization of traffic road maps play a key role in urban planning and traffic management. High-quality road maps can help decision makers make more informed planning decisions, improve the overall efficiency of the transportation system, and reduce traffic congestion and emissions. At present, the design of transportation road networks mainly relies on geographic information systems (GIS) and remote sensing technology, and manually plans and draws roads through satellite images and geographic data.

[0003] However, existing road map generation technology requires a lot of manual participation in the process, which makes it difficult to adapt to the rapidly changing needs of urban planning. It is inefficient when processing large-scale data sets and complex road networks, and its accuracy is limited by manual analysis capabilities. Summary of the invention

[0004] To this end, the embodiments of the present application provide a method, apparatus, device and storage medium for generating a road image, which improves the generation efficiency of the road image and improves the consistency of the generation results.

[0005] In a first aspect, the present application provides a method for generating a road image.

[0006] This application is achieved through the following technical solutions:

[0007] A method for generating a road image, comprising:

[0008] S1: Obtain the city road image to be mapped;

[0009] S2: converting the urban road image into a binary image;

[0010] S3: inputting the binary image into the trained generating agent to obtain the prediction coding information of the binary image;

[0011] S4: inputting the predicted coding information into the trained discriminant agent to obtain a connectivity evaluation result of the predicted coding information;

[0012] S5: When the connectivity evaluation result meets the preset requirements, the predicted coding information is subjected to inverse coding processing to obtain a predicted road network diagram;

[0013] S6: When the connectivity evaluation result does not meet the preset requirement, return to step S3 to regenerate the prediction coding information of the binary image.

[0014] In a preferred example of the present application, it can be further configured that before converting the urban road image into a binary image, the method further includes:

[0015] Using a lossless compression method to compress the urban road image;

[0016] Using image processing algorithms to crop or expand the compressed urban road images to unify the size of the compressed urban road images;

[0017] Convert city road images to the same image format.

[0018] In a preferred example of the present application, it can be further arranged that the generating agent and the discriminating agent are obtained by training based on the LLM model.

[0019] In a preferred example of the present application, it can be further configured that the LLM model is based on the LangChain architecture.

[0020] In a preferred example of the present application, it can be further set as follows:

[0021] The training process of the generating agent and the discriminating agent includes:

[0022] Obtain urban road images through public platforms;

[0023] Converting the urban road image into a binary image, and dividing the binary image into image blocks;

[0024] Encode the road connectivity of the image blocks according to a preset encoding format to obtain encoding information of all image blocks;

[0025] Inputting the image block and the coding information into a first LLM model to obtain predicted coding information of the image block, training the first LLM model based on the coding information and the predicted coding information, and using the first LLM model as a generated intelligent agent after its performance converges;

[0026] The predicted coding information and the corresponding connectivity labels are input into the second LLM model to obtain the connectivity evaluation result of the predicted coding information, and the second LLM model is trained based on the connectivity labels and the connectivity evaluation result. After the performance of the second LLM model converges, it is used as a discriminant agent.

[0027] In a preferred example of the present application, it can be further set as follows:

[0028] The road connectivity of the image blocks is encoded according to a preset encoding format to obtain encoding information of all image blocks, including:

[0029] Analyze the connectivity of any two of the four edges in the image block. If two of the edges are connected to each other, the connectivity of the two edges is set to a first value. If the two edges are not connected, the connectivity is set to a second value. The encoding information is represented as:

[0030] Code i,j ={x 12 ,x 13 ,x 14 ,x 23 ,x 34 ,x 24},

[0031] In the formula, Code i,j represents the encoding information of the image block in the i-th row and j-th column, x 12 Indicates the connectivity between the first and second edges, x 13 Indicates the connectivity between the first and third edges, x 14 Indicates the connectivity between the first and fourth edges, x 23 Indicates the connectivity between the second and third edges, x 34 Indicates the connectivity between the third and fourth edges, x 24 Indicates the connectivity between the 2nd and 4th edges.

[0032] In a second aspect, the present application provides a device for generating a road image.

[0033] This application is achieved through the following technical solutions:

[0034] A road image generation device, used to execute the road image generation method described in the first aspect, comprising:

[0035] An image acquisition module is used to acquire the urban road image to be mapped;

[0036] An image processing module, used for converting the urban road image into a binary image;

[0037] A generation module, used to obtain predictive coding information of road connectivity in the binary image;

[0038] A discriminant module, used to obtain a connectivity evaluation result about the predictive coding information;

[0039] The inverse coding module is used to perform inverse coding processing on the predicted coding information that meets the preset requirements to obtain a predicted road network diagram.

[0040] In a preferred example of the present application, it can be further set as follows:

[0041] The generation module and the discrimination module are obtained based on LLM model training.

[0042] On the third aspect, the present application is realized through the following technical solutions:

[0043] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of any one of the above-mentioned methods for generating a road image when executing the computer program.

[0044] In a fourth aspect, the present application provides a computer-readable storage medium.

[0045] This application is achieved through the following technical solutions:

[0046] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the computer program implements the steps of any one of the above-mentioned methods for generating a road image.

[0047] In summary, compared with the prior art, the technical solution provided in the embodiment of the present application has at least the following beneficial effects:

[0048] The present application obtains the city road image to be mapped; converts the city road image into a binary image; inputs the binary image into a trained generating agent to obtain the prediction coding information of the binary image; inputs the prediction coding information into a trained discriminating agent to obtain the evaluation result of the connectivity of the prediction coding information; when the evaluation result meets the preset requirements, the prediction coding information is inversely encoded to obtain a predicted road network diagram; when the evaluation result does not meet the preset requirements, the prediction coding information of the binary image is regenerated in the trained generating agent. The entire process of the road generation method of the present application starts from obtaining the city road image, through image binarization, agent prediction coding, evaluation of connectivity to the final generation of the predicted road network diagram, which is automatically completed without manual intervention. This greatly reduces the links and time of manual operation and improves the overall work efficiency. The entire process starts from obtaining the city road image, through image binarization, agent prediction coding, evaluation of connectivity to the final generation of the predicted road network diagram, which is automatically completed without manual intervention. This greatly reduces the links and time of manual operation and improves the overall work efficiency. It is possible to quickly find predictive coding information that meets the requirements, thereby improving the accuracy of road image generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1A schematic diagram of a process of generating a road image according to an embodiment of the present application;

[0050] Figure 2 A schematic diagram of the structure of a device for generating a road image provided in an embodiment of the present application;

[0051] Description of reference numerals:

[0052] Image acquisition module 10, image processing module 20, generation module 30, discrimination module 40, inverse encoding module 50. DETAILED DESCRIPTION

[0053] This specific embodiment is merely an explanation of the present application and is not a limitation of the present application. After reading this specification, those skilled in the art may make modifications to the present embodiment without any creative contribution as needed, but such modifications are protected by the patent law as long as they are within the scope of the claims of the present application.

[0054] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0055] In addition, the term "and / or" in this application is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this application, unless otherwise specified, generally indicates that the associated objects before and after are in an "or" relationship.

[0056] In this application, the terms "first", "second", etc. are used to distinguish identical or similar items with basically the same effects and functions. It should be understood that there is no logical or temporal dependency between "first", "second", and "nth", nor is there any limitation on quantity and execution order.

[0057] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0058] The embodiments of the present application are further described in detail below in conjunction with the drawings in the specification.

[0059] like Figure 1 A method for generating a road image provided by a first exemplary embodiment of the present application includes:

[0060] S1: Obtain the city road image to be mapped.

[0061] Specifically, the image of the city road to be mapped is collected by a camera device.

[0062] S2: Convert the urban road image into a binary image.

[0063] Specifically, in order to improve the processing accuracy of urban road images, the urban road images can be first filtered to remove noise in the urban road images. Usually, Gaussian filtering or median filtering can be selected. Then, the filtered urban road images are converted from color images to grayscale images for subsequent binarization processing. The Otsu algorithm is used to determine the binarization threshold, and the pixels in the grayscale image that are greater than the threshold are set to white (value 255), and the pixels that are less than the threshold are set to black (value 0), and the grayscale image is converted into a binary image, providing a basis for subsequent road recognition.

[0064] S3: Input the binary image into the trained generative agent to obtain the predicted coding information of the binary image.

[0065] After the binary image is input into the trained generative agent, the trained generative agent will segment the binary image into several image blocks, analyze the connectivity of the roads in each image block, and output the prediction coding information of the binary image. The prediction coding information is used to indicate the connectivity between the four edges in the image block.

[0066] Specifically, the encoding information can be expressed as: Code i,j ={x 12 ,x 13 ,x 14 ,x 23 ,x 34 ,x 24},

[0067] In the formula, Code i,j represents the encoding information of the sub-region of the i-th row and j-th column, x 12 Indicates the connectivity between the first and second edges, x 13 Indicates the connectivity between the first and third edges, x 14 Indicates the connectivity between the first and fourth edges, x 23 Indicates the connectivity between the second and third edges, x 34 Indicates the connectivity between the third and fourth edges, x24 Indicates the connectivity between the 2nd and 4th edges.

[0068] S4: Input the predicted coding information into the trained discriminant agent to obtain the connectivity evaluation result of the predicted coding information.

[0069] S5: When the connectivity assessment result meets the preset requirements, the predicted coded information is inversely encoded to obtain a predicted road network diagram.

[0070] S6: When the connectivity evaluation result does not meet the preset requirement, return to step S3 to regenerate the prediction coding information of the binary image.

[0071] In some preferred embodiments, before converting the urban road image into a binary image, the method further includes:

[0072] Use lossless compression method to compress the urban road image;

[0073] Using image processing algorithms to crop or expand the compressed urban road images to unify the size of the compressed urban road images;

[0074] Convert city road images of uniform size to the same image format.

[0075] Therefore, different city road images acquired by different image acquisition devices will differ in size and size. In order to provide images of uniform size for subsequent binarization processing, the city road images can be normalized first. Exemplarily, lossless compression of city road images can be achieved through JPEG 2000 lossless mode. Then the compressed city road images are unified into the same target size, for example, 512x512 pixels. If the size of the compressed city road image is larger than 512x512 pixels, the cv2.getRectSubPix() function of the Python-OpenCV library is called to crop it; if the size of the compressed city road image is smaller than 512x512 pixels, the edge of the compressed city road image is filled with zero pixels to make the image reach the target size. Finally, the Image.convert() function of the Python-PIL library is called to convert the image into a different format to convert the image into the target format. Preferably, the target format is .png format.

[0076] In some preferred embodiments, the generating agent and the discriminating agent are trained based on the LLM model. Both the generating agent and the discriminating agent are trained based on the LLM model and can learn complex road image features and rules from a large amount of historical data. The generating agent can accurately predict the coding information of the binary image, and the discriminating agent can accurately evaluate the connectivity of the predicted coding information, thereby ensuring that the generated road image is more accurate and in line with the actual situation in terms of structure and details.

[0077] Specifically, the LLM model is based on the LangChain architecture. Under the LangChain architecture, the problems handled by the large language model can be decomposed into multiple submodules. Each submodule has clear inputs, outputs, and functions, and can determine whether the set goals have been achieved. Different agents can be assigned to different submodules, and each agent can have its own unique prompt and called large language model. In addition, different agents can also be bound to different tools. Each agent can be equipped with a tool group, which contains tools that the agent may need to complete a specified task. These tools can be custom Python functions. The agent can select the appropriate tool function from the tool group to perform the task and monitor and observe the execution process.

[0078] The training process of generating agents and discriminating agents includes:

[0079] Obtain a large number of urban road images through the OSM public platform, which contain diverse regional morphology;

[0080] The Otsu algorithm is used to determine the binarization threshold of the urban road image. The urban road image is converted into a binary image according to the binarization threshold, and the binary image is divided into several image blocks. The size of the image block can be set according to the requirements.

[0081] Name the image blocks in the form of rows and columns, and assign the image blocks to the original binary image according to the sub-images;

[0082] Encode the road connectivity of the image blocks according to a preset encoding format to obtain encoding information of all image blocks, and use the encoding information to mark the image blocks;

[0083] The image block and the corresponding coding information are transmitted as input data to the first LLM model to obtain the predicted coding information of the image block. The loss function is defined based on the coding information and the predicted coding information. The gradient of the loss function for the first LLM model is calculated through the back propagation algorithm. The model parameters are updated using the optimizer. After the performance of the first LLM model converges, it is used as the generated intelligent agent.

[0084] The predicted coding information and the corresponding connectivity labels are input into the second LLM model to obtain the connectivity evaluation results of the predicted coding information. The loss function is defined based on the connectivity labels and the connectivity evaluation results. The gradient of the loss function to the model parameters is calculated through the back propagation algorithm, and the model parameters are updated using the optimizer. After the performance of the second LLM model converges, the model parameters are fixed and used as the discriminant agent.

[0085] The road connectivity of the image blocks is encoded according to a preset encoding format to obtain encoding information of all image blocks, including:

[0086] The four edges in the image block are marked as edge 1, edge 2, edge 3, and edge 4 in clockwise or counterclockwise direction.

[0087] Analyze the connectivity of any two of the four edges in the image block. If two of the edges are connected to each other, the connectivity of the two edges is set to the first value. If the two edges are not connected, the connectivity is set to the second value. The coded information is expressed as:

[0088] Code i,j ={x 12 ,x 13 ,x 14 ,x 23 ,x 34 ,x 24},

[0089] In the formula, Code i,j represents the encoding information of the image block in the i-th row and j-th column, x 12 Indicates the connectivity between the first and second edges, x 13 Indicates the connectivity between the first and third edges, x 14 Indicates the connectivity between the first and fourth edges, x 23 Indicates the connectivity between the second and third edges, x 34 Indicates the connectivity between the third and fourth edges, x 24 Indicates the connectivity between the 2nd and 4th edges.

[0090] The first value may be set to 1, and the second value may be set to 0.

[0091] Another embodiment of the present application further provides a road image generation device for executing the above method, such as Figure 2 As shown, the device comprises:

[0092] An image acquisition module 10 is used to acquire an image of a city road to be mapped;

[0093] An image processing module 20, used for converting the urban road image into a binary image;

[0094] A generating module 30, for obtaining predictive coding information of road connectivity in a binary image;

[0095] A determination module 40, for obtaining a connectivity evaluation result on the predictive coding information;

[0096] The inverse coding module 50 is used to perform inverse coding processing on the predicted coding information that meets the preset requirements to obtain a predicted road network diagram.

[0097] In some preferred embodiments, the image processing module 20 is further used to pre-process the acquired urban road image: compress the urban road image using a lossless compression method;

[0098] Using image processing algorithms to crop or expand the compressed urban road images to unify the size of the compressed urban road images;

[0099] Convert city road images to the same image format.

[0100] In some preferred embodiments, the generation module and the discrimination module are obtained based on LLM model training.

[0101] In some preferred embodiments, the LLM model is based on the LangChain architecture.

[0102] In some preferred embodiments, the generation module and the discrimination module are obtained based on the following training steps:

[0103] Obtain urban road images through public platforms;

[0104] Convert the urban road image into a binary image, and divide the binary image into image blocks;

[0105] Encode the road connectivity of the image blocks according to a preset encoding format to obtain encoding information of all image blocks;

[0106] Input the image block and the coding information into the first LLM model to obtain the predicted coding information of the image block, train the first LLM model based on the coding information and the predicted coding information, and use the first LLM model as a generation module after its performance converges;

[0107] The predicted coding information and the corresponding connectivity labels are input into the second LLM model to obtain the connectivity evaluation result of the predicted coding information. The second LLM model is trained based on the connectivity labels and the connectivity evaluation result. After the performance of the second LLM model converges, it is used as a discrimination module.

[0108] The specific definition of the road image generation device provided in this embodiment can be found in the embodiment of the road image generation method described above, and will not be repeated here. Each module in the above road image generation device can be implemented in whole or in part by software, hardware, or a combination thereof. Each of the above modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each of the above modules.

[0109] The embodiment of the present application provides a computer device, which may include a processor, a memory, a network interface and a database connected through a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the processor executes the steps of the method for generating a road image as in any of the above embodiments.

[0110] The working process, working details and technical effects of the computer device provided in this embodiment can be found in the embodiment of the method for generating a road image above, and will not be described in detail here.

[0111] The embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method for generating a road image in any of the above embodiments are implemented. The computer-readable storage medium refers to a carrier for storing data, which may include but is not limited to a floppy disk, an optical disk, a hard disk, a flash memory, a USB flash drive and / or a memory stick, etc. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0112] The working process, working details and technical effects of the computer-readable storage medium provided in this embodiment can be found in the embodiment of the method for generating a road image above, and will not be described in detail here.

[0113] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0114] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0115] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the system described in the present application is divided into different functional units or modules to complete all or part of the functions described above.

Claims

1. A method for generating a road image, characterized in that: include: S1: Obtain the city road image to be mapped; S2: converting the urban road image into a binary image; S3: inputting the binary image into the trained generating agent to obtain the prediction coding information of the binary image; S4: inputting the predicted coding information into the trained discriminant agent to obtain a connectivity evaluation result of the predicted coding information; S5: When the connectivity evaluation result meets the preset requirements, the predicted coding information is subjected to inverse coding processing to obtain a predicted road network diagram; S6: When the connectivity evaluation result does not meet the preset requirement, return to step S3 to regenerate the prediction coding information of the binary image.

2. The method for generating a road image according to claim 1, characterized in that: Before converting the urban road image into a binary image, the method further includes: Using a lossless compression method to compress the urban road image; Using image processing algorithms to crop or expand the compressed urban road images to unify the size of the compressed urban road images; Convert city road images to the same image format.

3. The method for generating a road image according to claim 1, characterized in that: The generating agent and the discriminating agent are obtained by training based on the LLM model.

4. The method for generating a road image according to claim 3, characterized in that: The LLM model is based on the LangChain architecture.

5. The method for generating a road image according to claim 3, characterized in that: The training process of the generating agent and the discriminating agent includes: Obtain urban road images through public platforms; Converting the urban road image into a binary image, and dividing the binary image into image blocks; Encode the road connectivity of the image blocks according to a preset encoding format to obtain encoding information of all image blocks; Inputting the image block and the coding information into a first LLM model to obtain predicted coding information of the image block, training the first LLM model based on the coding information and the predicted coding information, and using the first LLM model as a generated intelligent agent after its performance converges; The predicted coding information and the corresponding connectivity labels are input into the second LLM model to obtain the connectivity evaluation result of the predicted coding information, and the second LLM model is trained based on the connectivity labels and the connectivity evaluation result. After the performance of the second LLM model converges, it is used as a discriminant agent.

6. The method for generating a road image according to claim 5, characterized in that: The road connectivity of the image blocks is encoded according to a preset encoding format to obtain encoding information of all image blocks, including: Analyze the connectivity of any two of the four edges in the image block. If two of the edges are connected to each other, the connectivity of the two edges is set to a first value. If the two edges are not connected, the connectivity is set to a second value. The encoding information is represented as: Code i,j ={x 12 ,x 13 ,x 14 ,x 23 ,x 34 ,x 24 }, In the formula, Code i,j represents the encoding information of the image block in the i-th row and j-th column, x 12 Indicates the connectivity between the first and second edges, x 13 Indicates the connectivity between the first and third edges, x 14 Indicates the connectivity between the first and fourth edges, x 23 Indicates the connectivity between the second and third edges, x 34 Indicates the connectivity between the third and fourth edges, x 24 Indicates the connectivity between the 2nd and 4th edges.

7. A device for generating a road image, characterized in that: include: An image acquisition module is used to acquire the urban road image to be mapped; An image processing module, used for converting the urban road image into a binary image; A generation module, used to obtain predictive coding information of road connectivity in the binary image; A discriminant module, used to obtain a connectivity evaluation result about the predictive coding information; The inverse coding module is used to perform inverse coding processing on the predicted coding information that meets the preset requirements to obtain a predicted road network diagram.

8. The road image generation device according to claim 7, characterized in that: The generation module and the discrimination module are obtained based on LLM model training.

9. A computer device, characterized in that: The method comprises 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 according to any one of claims 1 to 7.

10. 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 steps of the method according to any one of claims 1 to 7 are implemented.

Images