Camera simulation method and device based on AI big data model

Abstract

The invention provides a camera simulation method and device based on an AI big data model, and the method comprises the steps: obtaining a camera parameter setting simulation camera adaptive to a test task demand, and calling the AI big data model to generate simulation test data; responding to the man-machine interaction operation to trigger the test system, and receiving simulation test data required by calling signals in the test task; acquiring simulation test data to be used when the test task is tested in real time, and executing a test action related to the test task on the simulation test data; and performing real-time operation processing on the selected simulation test data based on the to-be-used data to obtain real-time test result data of a calculation result obtained by comparing the simulation test data. The device comprises a cloud platform, a man-machine interaction module and a host module, the test environment is adapted through batch image data generated by an AI big data model, time sequence continuity and high real-time performance are achieved, and the actual use requirement of the image data test stage is met.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of image processing and application, and particularly relates to a camera simulation method and device based on an AI big data model. BACKGROUND

[0002] Since real cameras face multiple bottlenecks in acquiring data from real scenes, for example, professional cameras are limited in collecting image data in complex environments, and cannot ensure the time sequence continuity and real-time performance of image data, and are limited by camera configuration and continuous action of dynamic targets, image data is often discretely collected, resulting in loss of key frames of images or motion blur of typical targets in images, which is difficult to meet the strict requirements of various scenes on real-time response of data; in addition, due to the limitation of scene reproduction capability, there are inevitable disadvantages such as difficulty in deploying real cameras or long data acquisition period.

[0003] Camera simulation technology not only can make up for the above-mentioned disadvantages, but also can overcome the defects of time-consuming, labor-consuming and high cost of real cameras. Although the early camera simulation technology can rely on experts to deeply understand the imaging mechanism, it is difficult to adapt to complex and changeable scenes and cannot truly restore the imaging process, resulting in poor flexibility, rigid simulation effect and insufficient reality of generated images. With the rise of convolutional neural network and generative adversarial network, camera simulation technology can significantly improve the processing capacity and efficiency of images by optimizing feature matching, automatically extracting features and improving the adaptability of scenes through deep learning.

[0004] Patent publication document CN202092660U discloses a scene simulator suitable for TDI camera, which meets the requirement of accurately testing the dynamic imaging performance of the camera, and the purpose is to assist in verifying the dynamic imaging performance of the TDI camera, but it cannot be applied to on-site testing and verification of algorithms. And the existing camera simulation technology has limited generalization ability, which cannot meet the use requirements of different scenes, and cannot adaptively calculate different original images to quickly generate image data required for image processing. SUMMARY

[0005] The purpose of the present application is to solve the problems of time-consuming, labor-consuming, high cost, time discontinuity and insufficient real-time performance of test data when testing and verifying image processing algorithms based on real camera data from real scenes, configure a simulated camera according to the needs of image processing users, generate camera data that meets the real conditions, generate image data that meets the test requirements based on AI big data models and simulated cameras, and provide efficient assistance for image processing users in development, testing and verification. The batch image data generated by the AI big data model is adapted to the test environment of various types of image data, has time continuity and high real-time performance, and meets the actual use requirements of the image data test phase.

[0006] The technical solution provided by the present application is:

[0007] In one aspect, the present application provides a camera simulation method based on AI big data model, comprising:

[0008] Step one, simulate a camera by obtaining camera parameter settings that adapt to the test task requirements, and call AI big data model to generate simulation test data;

[0009] Step two, trigger the test test system in response to human-computer interaction operation, and receive simulation test data from real-time extraction and meeting the calling signal required in the test task;

[0010] Step three, real-time acquisition of simulation test data to be used during the test of the test task, and execution of test actions related to the test task;

[0011] Step four, real-time operation and processing of the selected simulation test data based on the task information corresponding to the test action of the to-be-used data, to obtain real-time test result data based on the calculation results obtained by comparing the simulation test data based on the test task.

[0012] Based on one aspect, the above-mentioned camera simulation method based on AI big data model comprises:

[0013] Extract camera data of various types of cameras and various camera interfaces, obtain test task requirements corresponding to the camera data and test verification targets in the test task;

[0014] Determine camera parameters by relying on camera data to adapt to test task requirements to determine test verification standards, and set parameters for simulation cameras to obtain various types of simulation cameras that match test task requirements;

[0015] Call AI big data model to generate simulation test data by customizing various types of simulation cameras.

[0016] Based on one aspect, the above-mentioned AI big data model-based camera simulation method, the step one of calling AI big data model to generate simulation test data, specifically includes:

[0017] The AI big data model is called to customize the data from the simulation camera request on demand, the corresponding simulation camera setting parameters of the test task requirement are selected to select the adaptive data generation link, and the data is called to generate simulation test data; wherein:

[0018] The data from the simulation camera includes:

[0019] The simulation data source stored in the simulation camera itself or accessed from the outside; at least containing any test data: test data read from the simulation camera local storage historical image data, test data required for recent test task test generation, test data obtained by accessing external storage medium;

[0020] The on-demand customization includes: according to the test task requirement, the parameters and various properties of the called data are flexibly adjusted, and a batch of simulation test data can be generated in the data generation link.

[0021] Based on one aspect, the above-mentioned AI big data model-based camera simulation method, the step one of calling AI big data model to generate simulation test data, specifically includes:

[0022] Read the setting parameters set in the simulation camera, call the AI big data model of Internet open source, and generate a batch of simulation test data meeting the test task by AI big data model and its data generation link according to the test task requirement and store it; Specifically:

[0023] The simulation data generation module in the simulation camera selects the data generation link according to the setting parameters;

[0024] The data generation link is used to automatically generate simulation test data based on the called data from the simulation test data; wherein, the simulation test data includes the source data stored in the simulation camera itself or accessed from the outside; the simulation test data is the image test data meeting the data output format and time sequence requirement of the test task, which is simulated real scene and real-time generated;

[0025] A batch of simulation test data is stored in the local, and the image test data is selected by the execution test task in the triggered test test system in real time.

[0026] Based on one aspect, the above-mentioned AI big data model-based camera simulation method, the step two includes:

[0027] A trigger signal is generated in response to a user operation or access to external data, triggering the test system including the host computer software to enter a working state;

[0028] The setting parameters of the test system are determined according to the requirements of the test task to be performed, and the parameters are set through the prompts of the host computer software;

[0029] The test data source required to meet the requirements of the test task to be performed is selected, and a calling signal that meets the timing requirements of the test data in the test task demand is determined;

[0030] The analog test data required for the test task to be performed is real-time called from the analog camera based on the calling signal.

[0031] On the one hand, the above-mentioned camera simulation method based on AI big data model comprises:

[0032] Real-time acquisition of continuous timing and high real-time analog test data for the test task to be performed, and execution of inspection actions, test actions and verification actions related to the test task.

[0033] On the one hand, the above-mentioned camera simulation method based on AI big data model comprises:

[0034] According to the test flow corresponding to the inspection action, test action and verification action, the selected continuous timing and high real-time analog test data are processed by an image processing algorithm to obtain test result data based on the selected analog test data and the algorithm processed test result data and stored in the camera.

[0035] On the other hand, the present application provides a camera simulation device based on AI big data model, comprising:

[0036] A cloud platform with AI big data model embedded;

[0037] A human-computer interaction module that acquires camera parameters adapted to the test task requirements and sets the current analog camera, and generates a trigger signal to trigger the test system;

[0038] A host module that communicates with the AI big data model and the human-computer interaction module, calls the AI big data model to generate analog test data at any time, receives a calling signal matched with the test task and real-time calls and selects the analog test data; the host module follows the test flow corresponding to the test action to real-time solve the selected analog test data.

[0039] Based on another aspect, the camera simulation device based on the AI big data model disclosed above further comprises:

[0040] An external input interface is connected to at least one of various existing cameras or external storage modules, so that the host module can obtain corresponding camera data or external storage data.

[0041] Based on another aspect, the camera simulation device based on the AI big data model disclosed above further comprises:

[0042] A test terminal is embedded with a test test system, and the test terminal tests or verifies the image processing algorithm based on the simulation test data called from the host module.

[0043] The beneficial effects of the present application are:

[0044] 1. The present application uses existing equipment and configures a simulation camera according to the needs of image processing users, quickly generates simulation test data meeting real conditions as camera data for user testing according to user testing needs, and provides efficient assistance for image processing users in development, testing and verification.

[0045] 2. The present application is based on AI big data model and simulation camera, fully utilizes the artificial intelligence advantage of AI big model to efficiently generate time-continuous and test requirement meeting image data, has time-continuous and high real-time advantages, and generates batch image data adapting to the test environment of various image data through AI big data model, meets the actual use needs of image data test stage, and can greatly improve the test verification efficiency.

[0046] 3. The present application can generate simulation test data of different formats, different interfaces and meeting scene verification needs according to user test verification and other test task needs, significantly improves the test efficiency of designed image processing algorithm and the ability of auxiliary positioning of designed image processing algorithm problems.

[0047] 4. The simulation camera of the present application integrates various common camera interfaces, has important auxiliary role for various cameras in product pre-design, or comparison of different camera interface performance, and does not need to purchase similar physical products; the simulation camera of the present application can generate simulation test data at any time in the late product development or algorithm test verification stage, facilitates users to perform various test verification and performance evaluation, and greatly reduces the development cost and development risk.

[0048] Additional aspects and advantages of the present application will be partially given in the following description, partially become obvious from the following description, or be understood through the practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0049] The above and / or additional aspects and advantages of the present application will become apparent and more readily appreciated from the following description, taken in conjunction with the following drawings of exemplary embodiments of the present application, wherein:

[0050] Figure 1 A flow chart of a camera simulation method based on an AI big data model is provided in the present application.

[0051] Figure 2 A workflow chart of a camera simulation method assisting in testing an image processing algorithm is provided in the present application.

[0052] Figure 3 A timing generation flow chart in the camera simulation method is provided in the present application.

[0053] Figure 4 A principle block diagram of a camera simulation device based on an AI big data model is provided in the present application.

[0054] Figure 5 A principle diagram of a host module in the camera simulation device is provided in the present application. DETAILED DESCRIPTION

[0055] Embodiments of the present application are described in detail below, which are exemplary and intended to explain the present application, and cannot be understood as a limitation of the present application.

[0056] Referring to Figure 1 The embodiment of the present application discloses a camera simulation method based on an AI big data model, which comprises the following steps one to four:

[0057] Step one, acquiring a camera parameter setting simulation camera adapted to the test task requirement, calling an AI big data model to generate simulation test data.

[0058] In specific application, when verifying the advantages and disadvantages of an algorithm, an image processing algorithm engineer often needs to support and verify with image data collected in a real environment, so as to effectively assist in positioning the problems of the algorithm in different scenes, the real-time performance of the algorithm, the accuracy of the algorithm, the success rate of the algorithm, and the faults in the restoration debugging process, etc. The traditional laboratory test method is to verify the feasibility and advantages and disadvantages of the algorithm by using a simulation software to simulate the original video data, but the performance of the algorithm in the actual running platform cannot be verified. Or, data collected by a camera is read into the data by a capture card for verification. The disadvantage of this method is that the collected image is not the data meeting the requirements of a real scene, and cannot accurately reflect the influence of camera noise, position movement, scene or weather change in an outdoor scene, and cannot completely verify the adaptability of the related algorithm under different conditions.

[0059] Therefore, the step one further comprises: (1) extracting camera data of various cameras and various camera interfaces, obtaining corresponding camera data and test verification target in the test task requirement corresponding to the test task; (2) determining camera parameters according to the test verification standard of the camera data to adapt to the test task requirement, setting parameters of the simulation camera to obtain various simulation cameras matching the test task requirement; wherein the test verification standard comprises resolution, frame rate, image type, data format, output interface, simulation data duration, noise distribution type, image special requirement superposition, special position insertion and simulation data source; (3) calling an AI big data model to generate simulation test data by customizing various simulation cameras to call shooting data.

[0060] On the basis of the above application, different interface types, different types of cameras and support of acquisition cards may be required for different engineering application algorithm development, which obviously increases the development cost. How to make the verification work of algorithm engineers simple and efficient, so that they can simulate and generate original data according to user requirements without real cameras, or search for original data from the Internet based on an AI big data model to train and generate the required data, or generate simulation signals according to the interface and timing format of real cameras according to the test requirements, and add noise and other interference test items.

[0061] Therefore, the step one further comprises:

[0062] a) calling an AI big data model to generate simulation test data, specifically comprising: calling an AI big data model to customize data requested from a simulation camera on demand, selecting an adaptive data generation link according to the relevant setting parameters of the simulation camera extracted according to the test task requirement, and calling data to generate simulation test data; wherein: (a1) the data requested from the simulation camera includes: simulation data sources stored in the simulation camera itself or accessed from the outside; at least containing any test data: test data read from the simulation camera local storage historical image data, test data required for recent test task test generation, test data obtained by accessing external storage medium; (a2) on-demand customization includes: flexibly adjusting various parameters and properties of the called data according to the test task requirement, and generating batch simulation test data in the data generation link.

[0063] b), calling an AI big data model to generate simulation test data, specifically comprising: reading the set parameters set in the simulation camera, calling an open-source AI big data model on the Internet, and generating batch simulation test data meeting the test task requirements from the AI big data model and its internal data generation link according to the test task requirements and storing them; specifically: (b1) the simulation data generation module in the simulation camera selects a data generation link according to the set parameters; (b2) the simulation test data is automatically generated based on the calling data in the simulation test data using the data generation link; wherein the simulation test data includes source data stored in the simulation camera itself or accessed from the outside; the simulation test data is simulation of real scenes, real-time generation of image test data meeting the data output format and timing requirements in the test task; (b3) the batch simulation test data is stored locally for the execution test task in the triggered test test system to select image test data in real time.

[0064] The above embodiment of the present application effectively solves the problems of time-consuming, labor-consuming, high cost, discontinuous timing and insufficient real-time test data of image test algorithm in test and verification according to the data obtained from real scenes by real cameras, configures the simulation camera according to the needs of image processing users, generates camera data meeting the real conditions, generates image data meeting the test requirements and having continuous timing based on the AI big data model and the simulation camera, and adapts various test environments of image data corresponding to batch image data generated by the AI big data model, which has continuous timing and high real-time, meets the actual use requirements of image data test phase, and provides efficient assistance for development, test and verification of image processing users in subsequent steps.

[0065] Step two, in response to the human-computer interaction operation triggering the test test system, receiving the simulation test data required by the calling signal meeting the test task from the real-time extraction.

[0066] Further, the above step two in the embodiment of the present application comprises: in response to the user operation or the access of external data constituting a trigger signal, triggering the test test system containing the host computer software to enter the working state; determining the set parameters of the test test system according to the requirements of the test task to be executed, and setting the parameters through the prompt of the host computer software; selecting the test test data source meeting the requirements of the test task to be executed, and determining the calling signal meeting the timing requirements of the test data in the test task requirements; based on the calling signal, real-time calling the simulation test data required by the test task to be executed from the simulation camera.

[0067] In the application, specifically includes: (21) the test system responds to human-computer interaction operation, triggers the test system containing host computer software to start entering the working state; wherein, the human-computer interaction operation includes the trigger signal operated by the test user or accessed from the outside;(22) the test user sets the simulation test parameters suitable for the simulation camera selected in the simulation test data generation stage;(23) the simulation test parameters are matched to select the test data source required by the test task from the external input link, and the calling signal matched with the test task and meeting the camera data timing requirement is obtained from the host computer software;(24) the required test data is obtained from the generated simulation test data and the external simulation test data, and the simulation test data is real-time called by matching the test action and the test process; wherein, the generated simulation test data includes the test test data generated and stored locally, the test test data obtained by AI big data model and stored locally; the external simulation test data is the test test data accessed from the outside.

[0068] Step three, real-time acquisition of simulation test data to be used in the test of the test task, and execution of the test action related to the test task.

[0069] Further, the above-mentioned step three in the embodiment of the application includes: real-time acquisition of simulation test data with continuous timing and high real-time for the test task to be executed, and execution of the test action related to the test task.

[0070] Step four, real-time operation processing of the selected simulation test data based on the task information corresponding to the test action to be used data reading, to obtain the real-time test result data based on the calculation result obtained by comparing the simulation test data of the test task.

[0071] Further, the above-mentioned step four in the embodiment of the application includes: according to the test process corresponding to the corresponding test action, test action and verification action, based on the selected continuous timing and high real-time simulation test data, the image is processed by image processing algorithm to obtain the test result data based on the selected simulation test data and the algorithm processing and stored in the camera.

[0072] Referring to Figure 2In practical applications, the process of the camera simulation auxiliary image processing user test verification of the embodiment of the present application is as follows: the camera simulation device is started, and then the host computer software is started. The user sets the parameters of the simulation camera (including camera type, image type, resolution, data bit width, frame frequency, data format, working frequency, external input link, etc.) in the host computer software. The simulation camera system selects the source of test data according to the external input link (including reading data from an external storage medium through a USB interface, reading historical test data from a local storage module, or generating test data by a local simulation software, or obtaining test data generated by a cloud AI big data model from the Internet). The external data download import is completed and can be saved in the local storage module. After the simulation camera system executes the simulation camera timing command, the test data is read into the cache in batches in advance, and the simulation signal is automatically output according to the user's selection or output according to the external trigger pulse. When the external trigger synchronization is selected, the simulation camera timing generation module generates the simulation camera timing with the external trigger pulse as the synchronization signal; when the automatic mode is selected, the simulation camera timing generation module automatically outputs continuous timing signals, and the user controls the start and stop of the timing signal in the host computer software. The test sequence output by the simulation camera system is connected to the user end by the corresponding simulation camera interface, and the corresponding test sequence is output to the display end for the user to view in real time, assisting the user to complete the test verification.

[0073] The implementation method for generating simulation test data by using an AI big data model is as follows: the user submits the required simulation camera test data type through the host computer software, including the test data format (black and white / color, visible light / infrared, 16bit / 8bit, PNG / RAW / JPG, etc.), the background of the image, the season weather of the image, the number of samples, and other requirements. After clicking the AI open source big model access option through the Internet, the open source big model generates the required data in real time based on the training model. After the generation is completed, the host computer software downloads all the test data from the cloud through the gigabit Ethernet interface, and the sample generation is completed.

[0074] Referring to Figure 3 and Figure 5In actual application, the workflow of the analog camera in the embodiment of the present application for generating analog real product, that is, batch analog test data timing signal, is as follows: after the host computer software is configured with the type of analog camera to be generated, the interface type, output frame frequency, working clock, data format, data packaging protocol, and the like of the camera are transmitted to the analog camera timing generation module 103, the analog camera timing generation module 103 starts the corresponding interface module, and the relevant modules output data according to the designed timing logic. Taking the generation of Cameralink Full transmission timing as an example, the parameters that need to be configured externally include pixel clock frequency, image resolution, data bit width 16bit / 14bit / 12bit / 10bit / 8bit, Full / Medium / Base mode, row valid / blanking period, field valid / blanking period, and the like. After the above parameters are specified by the user, the front end of the analog camera timing generation module 103 reads the cached test data from the PCIE interface according to the output timing from the memory of the front end, the test data is encoded by the FPGA logic according to the Cameralink timing, and the encoded data is finally output in the form of LVDS differential pair in the mode of parallel to serial.

[0075] Referring to Figure 4 The embodiment of the present application discloses a camera simulation device based on an AI big data model, which comprises a cloud platform, a man-machine interaction module 1 and a host module 2, wherein the cloud platform is embedded with an AI big data model, the man-machine interaction module 1 acquires camera parameters adapted to the test task requirements and sets the current analog camera, the man-machine interaction module 1 generates a trigger signal for triggering the test system, the host module 2 communicates with the AI big data model and the man-machine interaction module, and the host module 2 calls the AI big data model to generate analog test data at any time.

[0076] Referring to Figure 4The camera simulation device based on the AI big data model disclosed above in the embodiment of the application further comprises an external input interface, the external input interface comprises an external Ethernet input interface 3, an external USB data input interface 4, a simulation camera interface 5, and an external trigger interface 6. The external USB data input interface 4 is connected to an external storage module, so that the host module can correspondingly acquire the corresponding external storage data. The simulation camera interface 5 is connected to various existing cameras, so that the host module can correspondingly acquire the corresponding camera data. The user can use the camera simulation device to simulate test or test the designed image processing algorithm. Therefore, the camera simulation device based on the AI big data model disclosed above in the embodiment of the application further comprises a test terminal, the test terminal is embedded with a test test system, and the test terminal tests or verifies the image processing algorithm based on the simulation test data called from the host module. Therefore, the host module in the embodiment of the application further receives a calling signal matched with a test task and real-time calls and selects simulation test data. The host module real-time solves the selected simulation test data according to a test process corresponding to a test action, and matches the test terminal to test and verify the algorithm.

[0077] The display of the man-machine interaction module 1 displays the image and video data generated by the camera simulation device in real time, and assists the user in configuring various parameters of the camera to be simulated, controlling the generation and operation of the simulation time sequence, and the like. The host module 2 is the core processing unit of the camera simulation device, is responsible for reading and network accessing and downloading of external data, processing and displaying of simulation data, construction and generation of simulation camera data models, storage and accessing of data, data transmission and interaction between the simulation camera time sequence generation module, scheduling and process management of tasks, and the like. The external Ethernet input interface 3 is connected to the Internet, and accesses cloud data according to the task demand initiated by the host. The external USB data input interface 4 is responsible for reading data from external mobile hard disks, SD cards, U disks, and the like. The simulation camera output interface 5 is a support medium for different types of cameras, and selects a corresponding interface to output continuous test data according to the type of the camera to be simulated. The supported camera interfaces include SDI, Cameralink, HDMI, PAL, CoaxPress, GigE, USB3.0, LVDS, and the like. The external trigger interface 6 is a synchronous interface for external testing of the user, and outputs a single image or continuous video according to the frequency of a pulse when the user end sends a pulse model to request output of simulation test data.

[0078] The host module in the camera simulation device of the embodiment of the application comprises a core processor 101, a PCIE interface 102, a simulation camera timing generation module 103, an external trigger interface 104, a GPU display memory module 105, a cache DDR3 SDRAM 106, a gigabit Ethernet interface 107, a USB3.0 interface 108, a data storage module 109, a power supply module 110, a display HDMI interface 111, a mouse keyboard interface 112, and a clock management module 113. The core processor 101 adopts a high-performance multi-core CPU, is responsible for the management and task scheduling of the whole system, and is the core processing unit of the host. Specifically: (1) the PCIE interface 102 is a transmission interface realized based on a high-speed interconnection chip, is responsible for carrying batch simulation test data from the memory to the simulation camera timing generation module, and realizes real-time high-speed transmission of data. (2) the simulation camera timing generation module 103 adopts a high-performance Zynq UltraScale+MPSoC FPGA architecture, on the one hand, receives and caches batch data transmitted by the PCIE high-speed interface, and on the other hand, utilizes the high-speed processing logic of the FPGA to realize different camera interface protocols. Specifically, refer to Figure 5As shown, the FPGA logic circuit implemented sub-modules include: Cameralink transmission timing and interface protocol, HDMI transmission timing and interface protocol, SDI transmission timing and interface protocol, PAL transmission timing and interface protocol, CoaxPress transmission timing and interface protocol, GigE transmission timing and interface protocol, USB3.0 transmission timing and interface protocol, LVDS transmission timing and interface protocol. In addition, the analog camera timing generation module 103 is equipped with corresponding standard connectors or adapters according to different interfaces. Taking the Cameralink interface as an example, to cover the full bandwidth of the protocol, the Cameralink Full mode is adopted, which is composed of two MDR / SDR connectors, and can be downward compatible with Cameralink Medium and Cameralink Base, with a maximum clock frequency of 85MHz.(3) The external trigger interface 104 is connected to the analog camera timing generation module 103 through a coaxial cable. When the analog camera timing generation module 103 detects an external trigger pulse, it outputs test data.(4) The GPU memory module 105 provides operation support for the high-definition display of the camera simulation device and the high-speed operation of the internal simulation software.(5) The cache DDR3 SDRAM 106 provides support for the cache and operation processing of the high-speed CPU of the core processor 101.(6) The user connects to the Internet through the Gigabit Ethernet interface 107, interacts with the cloud open source AI big data model, and downloads test data generated according to user requirements in real time.(7) The USB3.0 interface 108 connects external storage media such as mobile hard drives, SD cards, and USB sticks, making it easy for users to import external test data.(8) The data storage module 109 uses a PCIE4.0 interface to support solid state drives of more than 2TB, mainly storing test data obtained externally and test data generated historically. In addition, an operating system runs on it to ensure that the entire simulation device can perform various tasks.(9) The power supply module 110 is responsible for power management services for the entire system, connected to 220V mains as input, to ensure stable operation of the system.(10) The display HDMI interface 111 can support the HDMI2.1 protocol version with a maximum of 4K high-definition display, serving as a human-computer interaction interface, responsible for displaying host computer software information, outputting image information, and displaying system management tasks.(11) The mouse and keyboard interface 112 uses a USB2.0 interface to connect the mouse and keyboard, providing interface services for user operation and input.(12) The clock management module 113 is responsible for providing working frequency for the analog camera timing generation module 103, according to the analog camera clock frequency and frame frequency determined by the user on the host computer, the system calculates the corresponding input frequency, output frequency, frequency multiplication / division mode, etc. Parameters are sent to the clock management module 113 to generate appropriate working frequency.

[0079] The simulation camera device in the embodiment of the present application integrates various common camera interfaces, and has an important auxiliary role for various cameras to perform investigation or compare the performance of different camera interfaces in the early product design, without the need to purchase similar physical products; the simulation camera device in the embodiment of the present application can also generate simulation test data at any time through the simulation camera device in the late product development or algorithm test and verification stage, so as to facilitate users to perform various test verification and performance evaluation, and greatly reduce the development cost and development risk.

[0080] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of various equivalent modifications or replacements within the technical range disclosed by the present application, and these modifications or replacements should be covered within the protection scope of the present application.

Claims

1. A camera simulation method based on an AI big data model, characterized in that, include: Step 1: Obtain camera parameter settings that are suitable for the experimental task requirements, simulate the camera, and call the AI ​​big data model to generate simulated test data; Step 2: Respond to the human-computer interaction operation to trigger the test system and receive simulated test data extracted in real time that meets the call signal required in the test task; Step 3: Acquire the simulated test data to be used during the test of the test task in real time, and perform test actions related to the test task. Step 4: Based on the task information corresponding to the test action read from the data to be used, perform real-time calculation and processing on the selected simulated test data to obtain real-time test result data based on the calculation results obtained by comparing the simulated test data with the test task.

2. The camera simulation method based on an AI big data model according to claim 1, characterized in that, Step one includes: Extract existing camera data from various types of cameras and various camera interfaces, and obtain the experimental task requirements corresponding to the camera data and the test verification target in the test task. Based on camera data, test and verification standards adapted to the experimental task requirements are determined to identify camera parameters. Then, the parameters of the simulated cameras are set to obtain various simulated cameras that match the experimental task requirements. The AI ​​big data model is invoked to retrieve shooting data from various customized simulated cameras to generate simulated test data.

3. A camera simulation method based on an AI big data model according to claim 1 or 2, characterized in that, Step one, which involves calling the AI ​​big data model to generate simulated test data, specifically includes: The AI ​​big data model is invoked to customize data from simulated camera requests on demand. Based on the relevant settings parameters extracted from the simulated camera corresponding to the experimental task requirements, an appropriate data generation link is selected, and the data is then used to generate simulated test data. Among these steps: Data retrieved from analog cameras includes: The simulated data source is either stored by the camera itself or accessed from external sources; it includes at least any of the following test data: test data read from historical image data stored locally by the simulated camera, test data generated for recent test tasks, and test data obtained by accessing external storage media. The on-demand customization includes: flexibly adjusting various parameters and attributes of the called data according to the needs of the test task, and generating batches of simulated test data in the data generation chain.

4. A camera simulation method based on an AI big data model according to claim 1 or 2, characterized in that, Step one, which involves calling the AI ​​big data model to generate simulated test data, specifically includes: The system reads the settings parameters set in the simulated camera, calls an open-source AI big data model from the internet, and generates and stores batch simulated test data that meets the experimental task requirements, based on the AI ​​big data model and its internal data generation chain. Specifically: The analog data generation module within the matching analog camera selects a data generation link according to the set parameters; The system employs a data generation link and automatically generates simulated test data based on data retrieved from the simulated test data. The simulated test data includes source data stored by the simulated camera itself or accessed from external sources. The simulated test data is image test data that simulates real-world scenarios and is generated in real time to meet the data output format and timing requirements of the experimental task. Batch simulation test data is stored locally, allowing the image test data to be selected in real time for the execution of test tasks in the triggered test system.

5. The camera simulation method based on an AI big data model according to claim 1, characterized in that, Step two includes: The system responds to user operations or accesses external data to generate a trigger signal, which then triggers the test system, which includes host computer software, to enter the working state. Determine the setting parameters of the test system according to the requirements of the test task to be performed, and set the parameters according to the prompts of the host computer software; Select the test data source required to meet the requirements of the test task to be performed, and determine the calling signal that meets the timing requirements of the test data in the test task requirements; Based on the call signal, the simulated test data required for the test task to be performed is retrieved in real time from the simulated camera.

6. The camera simulation method based on an AI big data model according to claim 1, characterized in that, Step three includes: It acquires continuous and highly real-time simulated test data for use in the test tasks to be performed, and executes inspection, testing and verification actions related to the test tasks.

7. The camera simulation method based on an AI big data model according to claim 1, characterized in that, Step four includes: Based on the corresponding test procedures for inspection, testing, and verification actions, and using the selected continuous time-series and high real-time simulation test data, the image processing algorithm is used to process the images, including real-time resolution, to obtain test result data based on the selected simulation test data and the algorithm-processed data, which is then stored in the camera.

8. A camera simulation device based on an AI big data model, characterized in that, include: The cloud platform has an embedded AI big data model. The human-computer interaction module acquires camera parameters that meet the requirements of the test task and sets the current simulation camera. The human-computer interaction module also generates a trigger signal to trigger the test system. The host module communicates with the AI ​​big data model and the human-computer interaction module. The host module calls the AI ​​big data model to generate simulated test data at any time. The host module receives call signals that match the test task and retrieves and selects the simulated test data in real time. The host module performs real-time calculations on the selected simulated test data in accordance with the test process corresponding to the test action.

9. A camera simulation device based on an AI big data model according to claim 8, characterized in that, Also includes: An external input interface is provided, which is connected to at least one of various existing cameras or external storage modules, so that the host module can obtain the corresponding camera data or external storage data.

10. A camera simulation device based on an AI big data model according to claim 8, characterized in that, Also includes: The test terminal has an embedded test system, which tests or verifies the image processing algorithm based on simulated test data called from the host module.

Images