Streaming media rearview mirror door opening early warning system based on AI algorithm and use method thereof

Abstract

The invention relates to the technical field of automobile rearview mirror door opening early warning, in particular to a streaming media rearview mirror door opening early warning system based on an AI algorithm and a using method thereof.The method comprises the following steps that firstly, a camera, an edge processor and a sensor are subjected to self-inspection, and preset optical calibration data are loaded; initializing an image preprocessing module, starting an optical compensation and environment adaptation algorithm, and correcting the collected initial image in real time; 2, all cameras synchronously collect video streams, a system carries out content analysis on a pixel point # imgabs1 # in the collected original image # imgabs0 #, and a key risk area is automatically identified; according to the importance of different regions, dynamic regional compression is carried out on the image, a key region adopts a low compression ratio # imgabs2 #, and an auxiliary region adopts a high compression ratio # imgabs3 #; according to the streaming media rearview mirror door opening early warning system based on the AI algorithm, aiming at the delay problem in the image compression and transmission process, the problems of image quality reduction and distortion are effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile rearview mirror door opening warning technology, and specifically to a streaming media rearview mirror door opening warning system based on an AI algorithm and a method for using the system. Background Art

[0002] The AI-based streaming media rearview mirror door opening warning system is an intelligent driving assistance system that uses the streaming media rearview mirror's camera to collect real-time image information from behind and to the sides of the vehicle. It then uses AI algorithms to analyze these images and identify pedestrians, bicycles, vehicles, and other targets approaching the vehicle. When the vehicle is parked and a collision risk is detected, the system alerts the driver and passengers through visual, auditory, or tactile means. This system can effectively reduce traffic accidents caused by improper door opening and improve driving and parking safety. In the process of compressing and transmitting images collected by the camera, the streaming media rearview mirror door opening warning system based on AI algorithm will compress the images in order to reduce delay, resulting in a decrease in image quality and the occurrence of distortion problems, which may cause the driver to misjudge the distance to the rear objects and increase driving risks. Therefore, to address the above problems, a streaming media rearview mirror door opening warning system based on AI algorithm and its use method are proposed. Summary of the Invention

[0003] The purpose of the present invention is to provide a streaming media rearview mirror door opening warning system based on AI algorithm and its use method, so as to solve the problem that in the process of image compression and transmission of the image collected by the camera, the streaming media rearview mirror door opening warning system based on AI algorithm will compress the image in order to reduce delay, resulting in a decrease in image quality and the occurrence of distortion, which may cause the driver to misjudge the distance to the rear object and increase driving risks.

[0004] To achieve the above object, the present invention provides the following technical solutions: The streaming media rearview mirror door opening warning system based on AI algorithm and its use method include the following steps: Step 1: Perform a self-test on the camera, edge processor, and sensor, and load the preset optical calibration data. Initialize the image preprocessing module, start the optical compensation and environment adaptation algorithms, and perform real-time correction on the acquired initial image; Step 2: All cameras collect video streams synchronously, and the system collects the original images Medium pixel Conduct content analysis to automatically identify key risk areas; According to the importance of different areas, the image is dynamically compressed by region, and the key areas use low compression rate , the auxiliary area uses a high compression ratio ; And obtain the current bandwidth of network transmission status in real time and the current latency of the processor load , based on the current bandwidth of the network transmission status obtained and the current latency of the processor load and the maximum available bandwidth , maximum allowed delay , calculate and automatically adjust the compression parameters , and compress the image to generate a compressed image ; Step 3: Compress the image Decode through a lightweight neural network model and output the decoded image , and call the super-resolution and denoising algorithms to restore details and perform geometric correction on the image to generate a reconstructed image ; The mathematical expression for decoding by the lightweight neural network model is:

[0005] Where, It is a lightweight neural network model; The calculation formula for super-resolution and denoising algorithms to restore details and correct geometry of images is:

[0006] Where, is the total variation regularization term of the image, is the regularization parameter; Step 4: Multiple images 、 , through the geometric transformation matrix H, real-time registration and splicing are performed to generate a global view, and the key areas are magnified and displayed, and then the risk rate of pedestrians or other obstacles in the processed image is calculated. Perform calculations; The geometric transformation matrix H is calculated as:

[0007] Where, The optimal geometric transformation matrix is ​​used to align and stitch multiple images. For the overlapping area; The risk rate of pedestrians or other obstacles in the image The calculation formula is:

[0008] Where, Features extracted by the target detection module, is the feature weight, is the sigmoid function; When there is a risk of pedestrians or other obstacles in the image, the system immediately reminds the driver through visual and voice alarm modules to ensure safe door opening.

[0009] Preferably, the method further comprises the following steps: Step 5: A1: When the system detects a pedestrian or other obstacle in the door opening area or blind spot, it triggers the warning mechanism and provides warning information in the form of graphics, text and sound on the instrument panel or display to ensure that the driver can react quickly; A2: Record actual operation data and warning events, including but not limited to image data, sensor data, and warning trigger conditions, and store the recorded data in the system operation data recording module; A3: Get old model parameters from the current AI model , obtain the newly collected data from the system operation data recording module , get the learning rate from the model training parameters , get the gradient of the loss function from the model training module ; A4: Calculate the updated model parameters using the formula , the updated model parameters The calculation formula is: ; A5: Update the model parameters Applied to AI models, it optimizes model performance and pushes optimized algorithms and model parameters to the system via OTA to ensure continuous system performance improvement. A6: The system continuously monitors actual operating data and warning events, regularly records data and updates models. Through continuous self-learning and optimization, the system can adapt to new driving environments and conditions.

[0010] Preferably, in step 2, the key area adopts a low compression rate and auxiliary areas using high compression ratios The determination rules are: .

[0011] Preferably, in step 2, the calculation formula of the compression parameter adjustment coefficient is: .

[0012] Preferably, the early warning mechanism includes the following triggering conditions: When the calculated risk rate When the preset threshold is exceeded, an early warning is triggered; When pedestrians or other obstacles are detected in the door opening area or blind spot, an early warning is triggered.

[0013] Preferably, the user interaction includes the following methods: Display warning information in graphic or text form on the instrument panel or display screen; Prompt the driver with warning information through voice; Attract the driver's attention through screen flashing or icon prompts.

[0014] Preferably, the data recording and model updating includes the following process: Regularly upload actual operation data and warning events to the cloud server; Analyze the data in the cloud server to generate new training data; Retrain the AI ​​model using new training data to generate updated model parameters; Push the updated model parameters to the system via OTA to complete the model update.

[0015] Preferably, it includes a camera module for collecting image data of the vehicle's surroundings; The edge processor module is used to perform preliminary processing on images captured by the camera, including image preprocessing, intelligent compression, and fast decoding; Sensor module, used to monitor environmental conditions, such as ambient light intensity, temperature, and rain; Image preprocessing module, used to perform real-time correction on the initial image captured by the camera, including optical correction and environment adaptation algorithms; AI compression module, used to dynamically compress images by region, using different compression rates based on region importance; A network transmission module is used to transmit the compressed image data to the central processing unit or display terminal via the vehicle-mounted high-speed network protocol; The decoding and image restoration module is used to decode the compressed image and call the super-resolution and denoising algorithms to restore details and perform geometric correction; Image fusion module, used to register and stitch images from different cameras in real time to generate a global view; The target detection module is used to monitor the processed images in real time and detect pedestrians or other obstacles in the door opening area, blind spots, and side and rear areas; A warning module, which alerts the driver through visual and voice alarm modules when risks are detected; User interaction module, used to provide information in the form of graphics, text and sound on the instrument panel or display; Data recording and storage module, used to record actual operation data and warning events, including image data, sensor data, and warning trigger conditions; Model optimization module, used to retrain the AI ​​model with newly collected data and generate updated model parameters; The OTA update module is used to push the optimized algorithm and model parameters to the system through wireless updates.

[0016] Preferably, a computer program is stored thereon, and when the computer program is executed by a processor, the steps in the method according to any one of claims 1 to 7 are executed.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. The AI-based streaming media rearview mirror door opening warning system in this invention addresses delays in image compression and transmission, effectively solving the problems of image quality degradation and distortion. The system uses a content-aware dynamic region-based compression algorithm to employ a low compression ratio for key areas, preserving more details, while employing a high compression ratio for auxiliary areas. This reduces the amount of data while maximizing the protection of key information. The system utilizes deep learning-driven super-resolution and denoising algorithms to restore details and perform geometric corrections on compressed images, further eliminating distortion that may be introduced by compression and ensuring image quality. The application of these technologies not only reduces image transmission delays but also improves image accuracy and reliability, preventing drivers from misjudging the distance to objects behind them due to image quality issues. This significantly improves driving safety, provides drivers with clearer and more accurate visual information, and effectively reduces the risk of collision when opening the door. 2. The AI-based streaming media rearview mirror door-opening warning system in this invention significantly improves driving safety through a multimodal warning mechanism. When the system detects a potential risk, it not only alerts the driver through visual cues, but also uses multiple methods such as voice alarms and screen flashing to ensure that the driver can quickly and accurately obtain warning information. This multimodal interaction method is particularly suitable for complex driving environments or when the driver's attention is distracted, and can effectively reduce the possibility of neglect or misunderstanding caused by a single warning method. 3. In the present invention, the system can continuously adjust and improve its own performance based on actual operating data. By regularly recording and uploading operating data and warning events, the system uses cloud servers to perform data analysis and model retraining to generate more optimized model parameters. These updated parameters are pushed to the system via OTA wireless updates to ensure that the system is always in the best performance state. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flow chart of the method for using the streaming media rearview mirror door opening warning system based on the AI ​​algorithm of the present invention; Figure 2 This is a system block diagram of the streaming media rearview mirror door opening warning system based on the AI ​​algorithm of the present invention. DETAILED DESCRIPTION

[0019] See also Figure 1-2 , the present invention provides a technical solution: The streaming media rearview mirror door opening warning system based on AI algorithm and its use method include the following steps: Step 1: Perform a self-test on the camera, edge processor, and sensor, and load the preset optical calibration data. Initialize the image preprocessing module, start the optical compensation and environment adaptation algorithms, and perform real-time correction on the collected initial image. This ensures that all modules of the system work normally when starting up and provides a high-quality foundation for initial image processing. Step 2: All cameras collect video streams synchronously, and the system collects the original images Medium pixel Conduct content analysis to automatically identify key risk areas; According to the importance of different areas, the image is dynamically compressed by region, and the key areas use low compression rate , the auxiliary area uses a high compression ratio ; And obtain the current bandwidth of network transmission status in real time and the current latency of the processor load , based on the current bandwidth of the network transmission status obtained and the current latency of the processor load and the maximum available bandwidth , maximum allowed delay , calculate and automatically adjust the compression parameters , and compress the image to generate a compressed image , protect image details in key areas, reduce data transmission volume, adapt to network fluctuations, and ensure low-latency transmission; Step 3: Compress the image Decode through a lightweight neural network model and output the decoded image , and call the super-resolution and denoising algorithms to restore details and perform geometric correction on the image to generate a reconstructed image ; The mathematical expression for decoding by the lightweight neural network model is:

[0020] Where, It is a lightweight neural network model; The calculation formula for super-resolution and denoising algorithms to restore details and correct geometry of images is:

[0021] Where, is the total variation regularization term of the image, Regularization parameters ensure fast transmission and high-quality restoration of image data, providing clear visual information; Step 4: Multiple images 、 , through the geometric transformation matrix H, real-time registration and splicing are performed to generate a global view, and the key areas are magnified and displayed, and then the risk rate of pedestrians or other obstacles in the processed image is calculated. Perform calculations; The geometric transformation matrix H is calculated as:

[0022] Where, The optimal geometric transformation matrix is ​​used to align and stitch multiple images. For the overlapping area; The risk rate of pedestrians or other obstacles in the image The calculation formula is:

[0023] Where, Features extracted by the target detection module, is the feature weight, is the sigmoid function; When there is a risk of pedestrians or other obstacles in the image, the system immediately reminds the driver through visual and voice alarm modules to ensure safe door opening, provide a comprehensive field of view and real-time risk warnings, and improve driving safety.

[0024] As a further optimized content of the present invention, the following steps are also included: Step 5: A1: When the system detects a pedestrian or other obstacle in the door opening area or blind spot, it triggers a warning mechanism and provides warning information in the form of graphics, text, and sound on the instrument panel or display screen to ensure that the driver can react quickly. Multi-modal warning ensures timely communication of information and improves driving safety. A2: Record actual operating data and warning events, including but not limited to image data, sensor data, and warning trigger conditions. Store the recorded data in the system operation data recording module, providing detailed data records to support subsequent analysis and optimization. A3: Get old model parameters from the current AI model , obtain the newly collected data from the system operation data recording module , get the learning rate from the model training parameters , get the gradient of the loss function from the model training module , ensuring accurate acquisition of data and parameters required for model updates; A4: Calculate the updated model parameters using the formula , the updated model parameters The calculation formula is: ,Continuously improve system performance through data-driven model updates; A5: Update the model parameters Applied to AI models, optimizing model performance, and pushing optimized algorithms and model parameters to the system via OTA to ensure continuous system performance improvements. Wireless updates ensure the system is always up to date and adaptable to new driving environments. A6: The system continuously monitors actual operating data and warning events, regularly records data and updates models. Through continuous self-learning and optimization, the system can adapt to new driving environments and conditions, forming a closed-loop optimization to ensure continuous improvement in system performance.

[0025] As a further optimization of the present invention, in step 2, the key area adopts a low compression rate and auxiliary areas using high compression ratios The determination rules are: , Key areas include vehicle door openings, blind spots, and side and rear areas. These areas have a significant impact on driving safety, so a low compression rate is used to preserve more image details. Auxiliary areas include areas around the vehicle that do not directly affect safety judgment. These areas have less impact on driving safety, so a high compression rate is used to reduce data transmission volume, maximize data transmission efficiency, and ensure image quality in key areas. As a further optimization of the present invention, in step 2, the calculation formula of the compression parameter adjustment coefficient is:

[0026] Dynamically adjust compression parameters to adapt to changes in network status and ensure transmission efficiency; As a further optimization of the present invention, the early warning mechanism includes the following triggering conditions: When the calculated risk rate When the preset threshold is exceeded, an early warning is triggered; When pedestrians or other obstacles are detected in the door opening area or blind spot, an early warning is triggered. Multiple conditions trigger to ensure timely warning and improve driving safety. As a further optimization of the present invention, user interaction includes the following methods: Display warning information in graphic or text form on the instrument panel or display screen; Prompt the driver with warning information through voice; Attract the driver's attention through screen flashing or icon prompts, and multi-form interaction ensures effective communication of information and enhances user experience; As a further optimization of the present invention, data recording and model updating include the following processes: Regularly upload actual operation data and warning events to the cloud server; Analyze the data in the cloud server to generate new training data; Retrain the AI ​​model using new training data to generate updated model parameters; Push the updated model parameters to the system via OTA to complete the model update. Cloud analysis and OTA updates ensure that the system is continuously optimized and adapts to the new environment. As a further optimized content of the present invention, wherein: a camera module is included for collecting image data of the vehicle's surrounding environment; The edge processor module is used to perform preliminary processing on images captured by the camera, including image preprocessing, intelligent compression, and fast decoding; Sensor module, used to monitor environmental conditions, such as ambient light intensity, temperature, and rain; Image preprocessing module, used to perform real-time correction on the initial image captured by the camera, including optical correction and environment adaptation algorithms; AI compression module, used to dynamically compress images by region, using different compression rates based on region importance; A network transmission module is used to transmit the compressed image data to the central processing unit or display terminal via the vehicle-mounted high-speed network protocol; The decoding and image restoration module is used to decode the compressed image and call the super-resolution and denoising algorithms to restore details and perform geometric correction; Image fusion module, used to register and stitch images from different cameras in real time to generate a global view; The target detection module is used to monitor the processed images in real time and detect pedestrians or other obstacles in the door opening area, blind spots, and side and rear areas; A warning module, which alerts the driver through visual and voice alarm modules when risks are detected; User interaction module, used to provide information in the form of graphics, text and sound on the instrument panel or display; Data recording and storage module, used to record actual operation data and warning events, including image data, sensor data, and warning trigger conditions; Model optimization module, used to retrain the AI ​​model with newly collected data and generate updated model parameters; OTA update module, used to push optimized algorithms and model parameters to the system through wireless updates. The modular design ensures comprehensive system functionality and is easy to maintain and upgrade. As a further optimization of the present invention, a computer program is stored thereon, and when the computer program is executed by a processor, the steps in any one of the methods of claims 1 to 7 are executed, and the software implementation ensures the flexibility and scalability of the system and facilitates functional upgrades.

[0027] This article uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only used to help understand the method of the present invention and its core ideas. The above is only a preferred implementation method of the present invention. It should be pointed out that due to the limitations of textual expression, there are objectively infinite specific structures. For ordinary technicians in this technical field, without departing from the principles of the present invention, they can make several improvements, modifications or changes, and can also combine the above technical features in an appropriate manner; these improvements, modifications, changes or combinations, or the direct application of the inventive concept and technical solution to other occasions without improvement, should be regarded as the scope of protection of the present invention.

Claims

1. The method for using the streaming media rearview mirror door opening warning system based on AI algorithm is characterized by: The following steps are involved: Step 1: Perform a self-test on the camera, edge processor, and sensor, and load the preset optical calibration data. Initialize the image preprocessing module, start the optical compensation and environment adaptation algorithms, and perform real-time correction on the acquired initial image; Step 2: All cameras collect video streams synchronously, and the system collects the original images Medium pixel Conduct content analysis to automatically identify key risk areas; According to the importance of different areas, the image is dynamically compressed by region, and the key areas use low compression rate , the auxiliary area uses a high compression ratio ; And obtain the current bandwidth of network transmission status in real time and the current latency of the processor load , based on the current bandwidth of the network transmission status obtained and the current latency of the processor load and the maximum available bandwidth , maximum allowed delay , calculate and automatically adjust the compression parameters , and compress the image to generate a compressed image ; Step 3: Compress the image Decode through a lightweight neural network model and output the decoded image , and call the super-resolution and denoising algorithms to restore details and perform geometric correction on the image to generate a reconstructed image ; The mathematical expression for decoding by the lightweight neural network model is: ； Where, It is a lightweight neural network model; The calculation formula for super-resolution and denoising algorithms to restore details and correct geometry of images is: ； Where, is the total variation regularization term of the image, is the regularization parameter; Step 4: Multiple images 、 , through the geometric transformation matrix H, real-time registration and splicing are performed to generate a global view, and the key areas are magnified and displayed, and then the risk rate of pedestrians or other obstacles in the processed image is calculated. Perform calculations; The geometric transformation matrix H is calculated as: ； Where, The optimal geometric transformation matrix is ​​used to align and stitch multiple images. For the overlapping area; The risk rate of pedestrians or other obstacles in the image The calculation formula is: ； Where, Features extracted by the target detection module, is the feature weight, is the sigmoid function; When there is a risk of pedestrians or other obstacles in the image, the system immediately reminds the driver through visual and voice alarm modules to ensure safe door opening.

2. The AI ​​algorithm-based streaming media rearview mirror door opening warning and its use method according to claim 1 are characterized by: The following steps are also included: Step 5: A1: When the system detects a pedestrian or other obstacle in the door opening area or blind spot, it triggers the warning mechanism and provides warning information in the form of graphics, text and sound on the instrument panel or display to ensure that the driver can react quickly; A2: Record actual operation data and warning events, including but not limited to image data, sensor data, and warning trigger conditions, and store the recorded data in the system operation data recording module; A3: Get old model parameters from the current AI model , obtain the newly collected data from the system operation data recording module , get the learning rate from the model training parameters , get the gradient of the loss function from the model training module ; A4: Calculate the updated model parameters using the formula , the updated model parameters The calculation formula is: ； A5: Update the model parameters Applied to AI models, it optimizes model performance and pushes optimized algorithms and model parameters to the system via OTA to ensure continuous system performance improvement. A6: The system continuously monitors actual operating data and warning events, regularly records data and updates models. Through continuous self-learning and optimization, the system can adapt to new driving environments and conditions.

3. The method for using the AI ​​algorithm-based streaming media rearview mirror door opening warning system according to claim 1 is characterized by: In step 2, low compression is used in the critical area and auxiliary areas using high compression ratios The determination rules are: 。 4. The method for using the AI ​​algorithm-based streaming media rearview mirror door opening warning system according to claim 1 is characterized by: In step 2, the calculation formula of the compression parameter adjustment coefficient is: 。 5. The method for using the AI ​​algorithm-based streaming media rearview mirror door opening warning system according to claim 1 is characterized by: The early warning mechanism includes the following triggering conditions: When the calculated risk rate When the preset threshold is exceeded, an early warning is triggered; When pedestrians or other obstacles are detected in the door opening area or blind spot, an early warning is triggered.

6. The method for using the AI ​​algorithm-based streaming media rearview mirror door opening warning system according to claim 1 is characterized by: The user interaction includes the following methods: Display warning information in graphic or text form on the instrument panel or display screen; Prompt the driver with warning information through voice; Attract the driver's attention through screen flashing or icon prompts.

7. The method for using the AI ​​algorithm-based streaming media rearview mirror door opening warning system according to claim 1 is characterized by: The data recording and model updating includes the following processes: Regularly upload actual operation data and warning events to the cloud server; Analyze the data in the cloud server to generate new training data; Retrain the AI ​​model using new training data to generate updated model parameters; Push the updated model parameters to the system via OTA to complete the model update.

8. The AI ​​algorithm-based streaming media rearview mirror door opening warning system according to any one of claims 1 to 7, characterized in that: including a camera module for collecting image data of the vehicle's surroundings; The edge processor module is used to perform preliminary processing on images captured by the camera, including image preprocessing, intelligent compression, and fast decoding; Sensor module, used to monitor environmental conditions, such as ambient light intensity, temperature, and rain; Image preprocessing module, used to perform real-time correction on the initial image captured by the camera, including optical correction and environment adaptation algorithms; AI compression module, used to dynamically compress images by region, using different compression rates based on region importance; A network transmission module is used to transmit the compressed image data to the central processing unit or display terminal via the vehicle-mounted high-speed network protocol; The decoding and image restoration module is used to decode the compressed image and call the super-resolution and denoising algorithms to restore details and perform geometric correction; Image fusion module, used to register and stitch images from different cameras in real time to generate a global view; The target detection module is used to monitor the processed images in real time and detect pedestrians or other obstacles in the door opening area, blind spots, and side and rear areas; A warning module, which alerts the driver through visual and voice alarm modules when risks are detected; User interaction module, used to provide information in the form of graphics, text and sound on the instrument panel or display; Data recording and storage module, used to record actual operation data and warning events, including image data, sensor data, and warning trigger conditions; Model optimization module, used to retrain the AI ​​model with newly collected data and generate updated model parameters; The OTA update module is used to push the optimized algorithm and model parameters to the system through wireless updates.

9. The AI ​​algorithm-based streaming media rearview mirror door opening warning system according to claim 1 is characterized by: A computer program is stored thereon, and when the computer program is executed by a processor, the steps in the method according to any one of claims 1 to 7 are executed.

Images