Automobile data recorder real-time image analysis method and system based on artificial intelligence

Abstract

The invention discloses an automobile data recorder real-time image analysis method and system based on artificial intelligence, and relates to the technical field of image data processing. The automobile data recorder real-time image analysis method based on artificial intelligence comprises the following steps of initial analysis, image fuzzy analysis and shutter adjustment triggering. Shutter influence adjustment judgment is carried out through the obtained initial judgment influence parameter, the shutter speed is adjusted based on the judgment result, then image fuzzy analysis is carried out on the automobile data recorder image to obtain the image quality analysis result, and finally the shutter speed adjustment measure is taken according to the image quality analysis result. Meanwhile, during image fuzzy analysis, if the shutter is triggered to influence adjustment judgment, shutter adjustment is triggered, so that the analysis stability of real-time images of the automobile data recorder is improved, and the problem that the driving state difference is not fully considered in the image shooting process of the automobile data recorder in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of image data processing, and particularly to a real-time image analysis method and system for a driving recorder based on artificial intelligence. Background Art

[0002] With the rapid development of artificial intelligence, computer vision, and autonomous driving technologies, real-time image analysis methods and systems for driving recorders based on artificial intelligence have been widely applied in the fields of intelligent transportation and driving safety. This technology uses deep learning and object detection algorithms to achieve automatic recognition and analysis of road environments, vehicles, pedestrians, and traffic signs, thereby providing functions such as driving assistance, accident warning, and evidence recording. In the face of complex traffic environments, this system improves driving safety and promotes the construction of an intelligent transportation system, laying a foundation for the development of future autonomous driving technologies.

[0003] Existing driving recorders mainly rely on traditional video recording functions and are difficult to achieve real-time intelligent analysis. Some advanced devices integrate advanced driver assistance systems that can perform functions such as lane departure warning and forward vehicle collision detection, but most are based on traditional algorithms and are easily affected by factors such as lighting and weather. In recent years, with the development of artificial intelligence and deep learning, driving recorders based on artificial intelligence can real-time identify pedestrians, vehicles, traffic signs, etc., and provide driving behavior analysis and safety warnings, greatly improving driving safety and the value of data utilization.

[0004] For example, the image data processing method, device, and equipment disclosed in the patent application with the publication number: CN114677440A include: the server obtains road information around the autonomous vehicle through the acquisition device on the autonomous vehicle. This road information may include road pictures collected from multiple perspectives at the same time. The server can input the received road information into a disparity model to calculate the depth picture corresponding to each road picture. The server can also input this depth picture into a preset recognition model to identify the target objects in the depth picture.

[0005] For example, a method for detecting the imaging quality of an in-vehicle camera disclosed in the invention patent announcement with the announcement number: CN114820623B includes: obtaining continuous M frames of imaging images collected by the in-vehicle camera during the test process and converting them into corresponding grayscale images; obtaining the texture complexity of each grayscale image, and screening out continuous N frames of tunnel junction images according to the grayscale mean and texture complexity of the grayscale image; obtaining the local uniformity of each tunnel junction image; obtaining the regional texture complexity of the tunnel entrance area and the tunnel interior area; obtaining the image quality of the tunnel junction image according to the local uniformity, area ratio, and regional texture complexity; evaluating the imaging quality of the in-vehicle camera according to the image quality difference and local uniformity difference between adjacent frame tunnel junction images, and the corresponding number of frames.

[0006] However, in the process of implementing the technical solution of the invention in the embodiments of the present application, it is found that the above technology has at least the following technical problems: In the prior art, since the shutter speed of the camera of the driving recorder is fixed, in inevitable objective situations such as high-speed movement, light change, and vehicle vibration, the exposure time is too long, and there is a problem that the driving recorder does not fully consider the difference in driving states during the process of taking pictures. Summary of the Invention

[0007] The embodiments of the present application provide a method and system for real-time image analysis of a driving recorder based on artificial intelligence, which solves the problem that the driving recorder does not fully consider the difference in driving states during the process of taking pictures in the prior art, and improves the analysis stability of the real-time image of the driving recorder.

[0008] The embodiments of the present application provide a method for real-time image analysis of a driving recorder based on artificial intelligence, including the following steps: performing a shutter influence adjustment judgment based on the obtained initial judgment influence parameters, and adjusting the shutter speed based on the judgment result; performing an image blur analysis on the driving recorder image to obtain an image quality analysis result, and taking a shutter speed adjustment measure according to the image quality analysis result; when performing the image blur analysis, if a shutter influence adjustment judgment is triggered, then trigger the shutter adjustment.

[0009] Further, the process of performing a shutter influence adjustment judgment based on the obtained initial judgment influence parameters and adjusting the shutter speed based on the judgment result is as follows: performing a difference operation on the initial judgment influence parameters and the corresponding initial judgment thresholds obtained from the preset database, and recording the initial judgment influence parameters corresponding to the difference operation result not less than 0 as the trigger judgment influence parameters; if there is only one trigger judgment influence parameter, input the initial deviation obtained based on the trigger judgment influence parameter and the corresponding initial judgment threshold into the first shutter adjustment mapping set to obtain the corresponding first shutter adjustment multiple, and perform a shutter speed adjustment operation based on the first shutter adjustment multiple and the shutter speed to obtain the first shutter speed, where the first shutter adjustment mapping set represents the mapping relationship between the initial deviation and the preset first shutter adjustment multiple; if there is more than one trigger judgment influence parameter, quantitatively evaluate the influence degree of the shutter speed on the image formation based on the initial deviation between the trigger judgment influence parameters and the corresponding initial judgment thresholds to obtain a shutter speed adjustment coefficient, input the real-time shutter speed adjustment coefficient into the second shutter adjustment mapping set to obtain the corresponding second shutter adjustment multiple, and perform a shutter speed adjustment operation on the shutter speed based on the second shutter adjustment multiple to obtain the first shutter speed, where the second shutter adjustment mapping set represents the mapping relationship between the shutter speed adjustment coefficient and the preset second shutter adjustment multiple; if there is no trigger judgment influence parameter, perform an image blur analysis.

[0010] Further, the specific process of obtaining the shutter speed adjustment coefficient is as follows: Number the obtained trigger judgment influence parameters according to the data type; Perform an initial deviation operation on the trigger judgment influence parameters and their initial judgment thresholds to obtain the corresponding initial deviation, and perform an influence weighting operation on the shutter speed adjustment weight obtained from the preset database and the obtained initial deviation to obtain the initial deviation influence value; Perform an equalization operation on the initial deviation influence value to obtain the shutter speed adjustment coefficient, and the shutter speed adjustment coefficient is used to quantify the influence degree of the shutter speed on the image imaging.

[0011] Further, the specific process of performing image blur analysis on the dash cam image to obtain the image quality analysis result and taking corresponding shutter speed adjustment measures according to the image quality analysis result is as follows: Perform image blur analysis on the real-time dash cam image to obtain the image quality analysis result, and the image quality analysis result includes global blur, local blur, and clear image; If the image quality analysis result is global blur, adjust the shutter speed according to the initial shutter adjustment ratio to obtain the second shutter speed. If the image quality analysis result obtained based on the second shutter speed is global blur, take the shutter iterative adjustment measure; If the image quality analysis result is local blur, perform an evaluation of the image blur influence effect based on the initial judgment influence parameters to obtain the image blur influence index, input the real-time image blur influence index into the image segmentation mapping set to obtain the corresponding image segmentation number, perform image segmentation according to the image segmentation number to obtain image fragments and perform image shutter adjustment, and the image segmentation mapping set represents the mapping relationship between the image blur influence index and the preset image segmentation number; If the image quality analysis result is a clear image, no additional processing is performed.

[0012] Further, the specific content of taking the shutter iterative adjustment measure is as follows: If the image quality analysis result obtained based on the second shutter speed is global blur, map the number of times of shutter speed adjustment to obtain the ratio reduction multiple, perform a ratio adjustment operation based on the ratio reduction multiple and the initial shutter adjustment ratio to obtain the adjustment ratio, and adjust the second shutter speed according to the adjustment ratio; When the number of times of adjusting the second shutter speed reaches the preset number and the image quality analysis result is still global blur, perform a voice cleaning prompt; When the number of times of adjusting the second shutter speed reaches the preset number and the image quality analysis result is local blur, perform a local blur analysis; When the number of times of adjusting the second shutter speed reaches the preset number and the image quality analysis result is a clear image, no additional processing is performed; After the voice cleaning prompt, it also includes a cleaning effectiveness judgment, and the specific process is as follows: After the voice cleaning prompt, if the image quality analysis result is global blur, perform a fault warning. If the image quality analysis result is local blur, perform a local blur analysis. If the image quality analysis result is a clear image, no additional processing is performed.

[0013] Further, the specific content of the image shutter adjustment is as follows: perform image blur analysis on the image fragments and classify them into clear fragments and blurred fragments, and perform an impact ratio operation on the counted number of clear fragments and the number of blurred fragments to obtain an image adjustment ratio; if the image blur impact index is less than the image impact threshold, input the image adjustment ratio into the shutter adjustment mapping set to obtain a shutter speed adjustment multiple, and perform a shutter adjustment operation on the shutter speed through the shutter speed adjustment multiple to obtain a third shutter speed. The shutter adjustment mapping set represents the mapping relationship between the image adjustment ratio and the preset shutter adjustment multiple; if the image blur impact index is not less than the image impact threshold, input the blur impact deviation obtained by performing a deviation operation on the image blur impact index and the image impact threshold into the ratio mapping set to obtain a ratio adjustment value, and input the adjusted image adjustment ratio obtained by performing an image adjustment ratio operation on the image adjustment ratio according to the ratio adjustment value into the shutter adjustment mapping set to obtain the corresponding shutter speed adjustment multiple to adjust the shutter speed to obtain a third shutter speed. The ratio mapping set represents the mapping relationship between the blur impact deviation and the ratio adjustment value.

[0014] Further, the method for obtaining the image blur impact index is as follows: obtain the initial judgment impact parameter in real time and perform a unitless processing on the initial judgment impact parameter; perform a difference degree operation on the initial judgment impact parameter and the corresponding initial judgment threshold to obtain a difference value; obtain the image shutter impact weight from a preset database and perform a weighting operation with the image blur impact index to obtain an image blur impact degree value; perform a balancing operation on the total number of data types of the image blur impact degree value and the initial judgment impact parameter to obtain the image blur impact index, and the image blur impact index is used to quantify the distance degree between the initial judgment impact parameter and the initial judgment threshold.

[0015] Further, when performing image blur analysis, if a shutter impact adjustment judgment is triggered, trigger shutter adjustment. The specific content is as follows: input the counted number of shutter speed adjustment times into the real-time adjustment mapping set to obtain a corresponding shutter adjustment impact value. The real-time adjustment mapping set represents a set of mapping relationships between the counted number of shutter speed adjustment times and the preset shutter adjustment impact value; during the process of performing image blur analysis, if a shutter impact adjustment judgment is triggered and there is only one trigger judgment impact parameter, perform an initial adjustment impact operation on the initial deviation according to the shutter adjustment impact value to obtain an adjusted initial deviation; if there is more than one trigger judgment impact parameter, perform a coefficient adjustment operation on the shutter speed adjustment coefficient according to the shutter adjustment impact value to obtain an adjusted shutter speed adjustment coefficient.

[0016] Further, during the process of image blur analysis, it also includes a shutter adjustment feasibility analysis, and the specific content is as follows: When making the first judgment on shutter impact adjustment, obtain the preset shutter adjustment analysis time period. If the number of times of making the shutter impact adjustment judgment reaches the preset judgment number within the shutter adjustment analysis time period, then capture an image based on the first shutter speed obtained from the last shutter impact adjustment judgment without making shutter adjustment judgment and image blur analysis until the end of the shutter adjustment analysis time period.

[0017] The embodiment of the present application provides a real-time image analysis system for a driving recorder based on artificial intelligence, including: an initial analysis module, an image blur analysis module, and a trigger shutter adjustment module; wherein, the initial analysis module is used to make a shutter impact adjustment judgment based on the obtained initial judgment impact parameters and adjust the shutter speed based on the judgment result; the image blur analysis module is used to perform image blur analysis on the driving recorder image to obtain an image quality analysis result and take shutter speed adjustment measures according to the image quality analysis result; the trigger shutter adjustment module is used to perform trigger shutter adjustment when triggering a shutter impact adjustment judgment during image blur analysis.

[0018] One or more technical solutions provided in the embodiment of the present application have at least the following technical effects or advantages: 1. By making a shutter impact adjustment judgment based on the obtained initial judgment impact parameters, adjusting the shutter speed based on the judgment result, then performing image blur analysis on the driving recorder image to obtain an image quality analysis result, and finally taking shutter speed adjustment measures according to the image quality analysis result, and at the same time, when performing image blur analysis, if a shutter impact adjustment judgment is triggered, then perform trigger shutter adjustment, so as to adjust the shutter speed more accurately, and thus improve the analysis stability of the real-time image of the driving recorder, effectively solving the problem that the driving state difference is not fully considered in the prior art during the process of the driving recorder capturing images.

[0019] 2. By numbering the obtained trigger judgment impact parameters according to the data type, then performing an initial deviation operation on the trigger judgment impact parameters and their initial judgment thresholds to obtain the corresponding initial deviation, then performing an influence weighting operation on the shutter speed adjustment weight obtained from the preset database and the obtained initial deviation to obtain an initial deviation influence value, and finally performing an equilibrium operation on the initial deviation influence value to obtain a shutter speed adjustment coefficient, so as to more accurately quantify the influence degree of the shutter speed on image imaging, and then timely adjust the shutter speed to improve the image quality.

[0020] 3. By obtaining the initial judgment influence parameters in real time, de - unitizing the initial judgment influence parameters, then calculating the difference value by calculating the degree of difference between the initial judgment influence parameters and the corresponding initial judgment thresholds, then obtaining the image shutter influence weight from the preset database, and performing a weighted operation with the image blur influence index to obtain the image blur influence degree value, and finally performing an equilibrium operation on the total number of data types of the image blur influence degree value and the initial judgment influence parameters to obtain the image blur influence index, thereby more accurately quantifying the influence degree of the initial judgment influence parameters on the real - time image imaging, and further improving the stability of the image imaging. Brief Description of the Drawings

[0021] Figure 1 It is a flowchart of a real - time image analysis method for a driving recorder based on artificial intelligence provided by an embodiment of the present application. Detailed Embodiment

[0022] In an embodiment of the present application, by providing a real - time image analysis method and system for a driving recorder based on artificial intelligence, the problem in the prior art that the difference in driving states is not fully considered during the process of the driving recorder shooting images is solved. The difference operation is performed between the initial judgment influence parameters and the corresponding initial judgment thresholds obtained from the preset database, and the initial judgment influence parameters corresponding to the result of the difference operation not less than 0 are recorded as the trigger judgment influence parameters. If there is only one trigger judgment influence parameter, the initial deviation obtained based on the trigger judgment influence parameter and the corresponding initial judgment threshold is input into the first shutter adjustment mapping set to obtain the corresponding first shutter adjustment multiple, and then the first shutter speed is obtained by performing a shutter speed adjustment operation based on the first shutter adjustment multiple and the shutter speed. If there are more than one trigger judgment influence parameters, the influence degree of the shutter speed on the image imaging is quantitatively evaluated based on the initial deviation between the trigger judgment influence parameters and the corresponding initial judgment thresholds to obtain the shutter speed adjustment coefficient. Then, the real - time shutter speed adjustment coefficient is input into the second shutter adjustment mapping set to obtain the corresponding second shutter adjustment multiple. Then, the first shutter speed is obtained by performing a shutter speed adjustment operation on the shutter speed based on the second shutter adjustment multiple. Then, the image blur analysis of the driving recorder image is performed to obtain the image quality analysis result. Finally, shutter speed adjustment measures are taken according to the image quality analysis result. At the same time, when performing the image blur analysis, if the trigger shutter influence adjustment judgment is triggered, the trigger shutter adjustment is performed, thereby improving the analysis stability of the real - time image of the driving recorder.

[0023] The technical solution in the embodiment of the present application aims to solve the above - mentioned problem that the difference in driving states is not fully considered during the process of the driving recorder shooting images. The general idea is as follows: Perform shutter impact adjustment judgment based on the obtained initial judgment impact parameters, adjust the shutter speed based on the judgment result, then perform image blur analysis on the dashcam image to obtain an image quality analysis result, and finally take shutter speed adjustment measures according to the image quality analysis result. At the same time, when performing image blur analysis, if the shutter impact adjustment judgment is triggered, trigger shutter adjustment, achieving the effect of improving the stability of real-time image analysis of the dashcam.

[0024] To better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings of the specification and specific embodiments.

[0025] As Figure 1 shown, it is a flowchart of a real-time image analysis method for a dashcam based on artificial intelligence provided by an embodiment of the present application. The method includes the following steps: perform shutter impact adjustment judgment based on the obtained initial judgment impact parameters, and adjust the shutter speed based on the judgment result; perform image blur analysis on the dashcam image to obtain an image quality analysis result, and take shutter speed adjustment measures according to the image quality analysis result; when performing image blur analysis, if the shutter impact adjustment judgment is triggered, perform trigger shutter adjustment.

[0026] In this embodiment, when the vehicle is driving, especially at high speed, the shutter speed of the dashcam is set fixed, which may cause long exposure, resulting in motion blur in the captured images, thereby reducing the image quality. By performing shutter impact adjustment judgment, real-time adjustment of the dashcam camera is achieved. At the same time, by performing image blur analysis on the dashcam, the impact of the shutter speed on the image imaging quality is reflected, thereby realizing the adjustment of the shutter speed based on the real-time analysis result of the image, and further improving the image quality recorded by the dashcam and ensuring the stability of the image quality recorded by the real-time analysis dashcam.

[0027] It should be added that the image blur analysis is realized through an artificial intelligence large model (such as a deep learning model). The specific steps are as follows: Step 1, data preparation. Collect a large amount of labeled image data, including clear images, globally blurred images, and locally blurred images. And in order to train an effective deep learning model, the data set needs to have diversity, including different types of blur (such as motion blur, focus blur, light blur, etc.) and clear images; and each image needs to have a label indicating whether it is blurred and distinguishing whether it is globally blurred or locally blurred.

[0028] Step 2, Feature Extraction: Through feature extraction, the deep learning model can identify the distribution of blurred regions in the image. Global blur usually manifests as the overall loss of sharpness in the image, while local blur appears as blurred parts in some regions of the image with the other regions remaining clear. For example, the gradient in local regions of the gradient image is small, and global blur will cause the overall gradient of the image to decrease. By performing a Fast Fourier Transform (FFT) to convert to the frequency domain, blurred images usually show a lack of low-frequency information, especially in the case of global blur. The edge information of the blurred image is unclear, and local blur may cause some edges to be missing or unclear.

[0029] Step 3, Model Selection: Convolutional neural networks are used to handle image classification problems and can automatically extract the spatial features of images. For blurred images, convolutional neural networks can extract local and global features through multiple convolutional and pooling layers, thereby determining whether the image is blurred and further distinguishing the type of blur. U-Net: For image segmentation tasks, the U-Net model is used to locate the blurred regions in the image and is suitable for distinguishing local and global blur. Combining convolutional neural networks and long short-term memory networks helps to achieve temporal analysis of blurred images and is suitable for blur analysis in dynamic images or video streams to help distinguish the type of blur.

[0030] Step 4, Model Training: Using labeled training data, train the model to predict whether the image is blurred and distinguish between global and local blur. During the training process, the model is continuously optimized by adjusting the parameters through backpropagation to minimize the error. To avoid overfitting, the training data is often enhanced by rotation, cropping, flipping, brightness adjustment, etc. to simulate blurred situations in different environments.

[0031] Step 5, Model Evaluation: Use the cross-validation method to evaluate the model performance and measure its performance in determining whether the image is blurred and distinguishing the type of blur. Evaluate the model's recognition ability for different types of blur (global and local blur) through a confusion matrix to ensure that the model can correctly distinguish the type of blur.

[0032] Step 6, Inference Phase: Input the image to be determined into the model. The model will analyze the features of the image to determine whether the image is blurred and, based on the distribution of the blurred regions, determine whether it is global or local blur. The output of the model can be a binary classification result (whether it is blurred) and a label (global or local blur), or a more precise pixel-level output showing the blurred regions in the image.

[0033] Further, perform shutter impact adjustment judgment based on the obtained initial judgment impact parameter, and adjust the shutter speed based on the judgment result. The specific process is as follows: Perform a difference operation between the initial judgment impact parameter and the corresponding initial judgment threshold obtained from the preset database, and record the initial judgment impact parameter corresponding to the result of the difference operation being not less than 0 as the trigger judgment impact parameter; If there is only one trigger judgment impact parameter, input the initial deviation obtained based on the trigger judgment impact parameter and the corresponding initial judgment threshold into the first shutter adjustment mapping set to obtain the corresponding first shutter adjustment multiple, and perform a shutter speed adjustment operation based on the first shutter adjustment multiple and the shutter speed to obtain the first shutter speed. The first shutter adjustment mapping set represents the mapping relationship between the initial deviation and the preset first shutter adjustment multiple; If there is more than one trigger judgment impact parameter, quantify and evaluate the impact of the shutter speed on image imaging based on the initial deviation between the trigger judgment impact parameter and the corresponding initial judgment threshold to obtain the shutter speed adjustment coefficient, input the real-time shutter speed adjustment coefficient into the second shutter adjustment mapping set to obtain the corresponding second shutter adjustment multiple, and perform a shutter speed adjustment operation on the shutter speed based on the second shutter adjustment multiple to obtain the first shutter speed. The second shutter adjustment mapping set represents the mapping relationship between the shutter speed adjustment coefficient and the preset second shutter adjustment multiple; If there is no trigger judgment impact parameter, perform image blur analysis.

[0034] In this embodiment, the initial judgment threshold is set by the preset staff according to the specific model and specific application situation of the driving recorder, and the initial judgment threshold represents the maximum initial judgment impact parameter corresponding to the driving recorder being able to obtain a clear image based on the current shutter speed; The difference operation means performing a subtraction operation between the initial judgment impact parameter and the corresponding initial judgment threshold; Through the above two different adjustment methods for obtaining the first shutter speed, the adjustment of the shutter speed is refined more precisely, ensuring that different adjustment methods are adopted for the adjustment of the shutter speed according to the real-time actual situation, thereby achieving a more accurate adjustment of the shutter speed, and further improving the quality of the real-time image of the driving recorder and ensuring the stability of the image quality recorded by the real-time analysis driving recorder.

[0035] It should be added that the initial judgment influence parameters include light intensity, vehicle speed, number of pedestrians, and humidity. Taking the above data as an example, the light intensity is obtained by directly integrating an ambient light sensor (such as chips like TSL2561 and VEML7700 of Austria Micro Systems) into the driving recorder, the vehicle speed is obtained by directly reading the real-time vehicle speed through the vehicle CAN (Controller Area Network) bus (the driving recorder needs to support connecting to OBD, i.e., On-Board Diagnostics), the number of pedestrians is obtained by real-time detecting the pedestrians around the vehicle based on object detection models such as YOLOv5 and Faster R-CNN, and the humidity is obtained by a humidity sensor installed near the outside rearview mirror or the front windshield of the vehicle.

[0036] Further, the specific process of obtaining the shutter speed adjustment coefficient is as follows: Number the obtained trigger judgment influence parameters according to the data type; perform an initial deviation operation on the trigger judgment influence parameters and their initial judgment thresholds to obtain the corresponding initial deviation, and perform an influence weighting operation on the shutter speed adjustment weight obtained from the preset database and the obtained initial deviation to obtain the initial deviation influence value; perform an equilibrium operation on the initial deviation influence value to obtain the shutter speed adjustment coefficient, and the shutter speed adjustment coefficient is used to quantify the influence degree of the shutter speed on the image imaging.

[0037] The specific limiting expression of the shutter speed adjustment coefficient is as follows: ; In the formula, j represents the data type number of the trigger judgment influence parameter, , represents the total number of data types of the trigger judgment influence parameter, represents the j-th type of trigger judgment influence parameter, represents the j-th type of initial judgment threshold, represents the shutter speed adjustment weight corresponding to the j-th type of trigger judgment influence parameter, and Y represents the shutter speed adjustment coefficient.

[0038] In this embodiment, the algorithm combines the trigger judgment influence parameter, the initial judgment threshold, and the shutter speed adjustment weight to comprehensively analyze and obtain the shutter speed adjustment coefficient. In the formula, when the trigger judgment influence parameter is greater than the initial judgment threshold, it indicates that the current trigger judgment influence parameter has a higher influence on the imaging of the dash cam based on the current shutter speed, and the corresponding shutter speed adjustment coefficient is larger. It should be noted that when the trigger judgment influence parameter is obtained, it is necessary to re - sort and number the trigger judgment influence parameter according to the data type; by performing a mean operation on the sum of the products of the initial deviations of all trigger judgment influence parameters and the shutter speed adjustment weight respectively, the obtained shutter speed adjustment coefficient more accurately and comprehensively quantifies the influence of the shutter speed on the imaging of the dash cam, so as to adjust the shutter speed in real - time, and then obtain a higher - quality real - time image of the dash cam.

[0039] Specifically, the shutter speed adjustment weight is obtained from a preset database. The shutter speed adjustment weight represents the influence degree of the trigger judgment influence parameter on the shutter speed adjustment coefficient. Each trigger judgment influence parameter and the shutter speed adjustment weight have a unique mapping relationship, and the value range is between 0 and 1. For example, a mapping set of the trigger judgment influence parameter and the preset shutter speed adjustment weight is constructed. The real - time trigger judgment influence parameter is input into the mapping set to obtain the corresponding shutter speed adjustment weight, which represents the influence degree of the trigger judgment influence parameter on the shutter speed adjustment coefficient, and the sum of all shutter speed adjustment weights is 1.

[0040] Furthermore, image blur analysis is performed on the dash cam image to obtain the image quality analysis result, and corresponding shutter speed adjustment measures are taken according to the image quality analysis result. The specific process is as follows: Artificial intelligence is used to perform image blur analysis on the real - time image of the dash cam to obtain the image quality analysis result. The image quality analysis result includes global blur, local blur, and clear image. If the image quality analysis result is global blur, the shutter speed is adjusted according to the initial shutter adjustment ratio to obtain the second shutter speed. If the image quality analysis result based on the second shutter speed is global blur, shutter iteration adjustment measures are taken. If the image quality analysis result is local blur, the image blur influence effect is evaluated based on the initial judgment influence parameter to obtain the image blur influence index. The real - time image blur influence index is input into the image segmentation mapping set to obtain the corresponding image segmentation number. Image segmentation is performed according to the image segmentation number to obtain image fragments and image shutter adjustment is performed. The image segmentation mapping set represents the mapping relationship between the image blur influence index and the preset image segmentation number. If the image quality analysis result is a clear image, no additional processing is performed.

[0041] In this embodiment, the initial shutter adjustment ratio is preset by a preset staff according to the model of the driving recorder camera and its specific application scenarios, and is used to adjust the shutter speed for the first time based on the globally blurred image; and by processing the real-time images of the driving recorder with blurred situations in different cases, it is beneficial to improve the real-time performance of image processing, reduce the delay of real-time image analysis, and ensure the timeliness of the images recorded by the driving recorder and the improvement of image quality.

[0042] Further, the specific content of the shutter iterative adjustment measure is as follows: If the image quality analysis result based on the second shutter speed is globally blurred, the ratio reduction multiple is mapped according to the number of shutter speed adjustments, and the adjustment ratio is obtained by performing a ratio adjustment operation on the ratio reduction multiple and the initial shutter adjustment ratio, and the second shutter speed is adjusted according to the adjustment ratio; when the number of adjustments of the second shutter speed reaches the preset number and the image quality analysis result is still globally blurred, a voice cleaning prompt is made; when the number of adjustments of the second shutter speed reaches the preset number and the image quality analysis result is locally blurred, local blur analysis is performed; when the number of adjustments of the second shutter speed reaches the preset number and the image quality analysis result is a clear image, no additional processing is performed. The voice cleaning prompt means that the driver is prompted to clean the camera of the driving recorder; after the voice cleaning prompt, it also includes a cleaning effectiveness judgment. The specific process is as follows: After the voice cleaning prompt, if the image quality analysis result is globally blurred, a fault warning is made; if the image quality analysis result is locally blurred, local blur analysis is performed; if the image quality analysis result is a clear image, no additional processing is performed. The fault warning means that the driver is prompted that the driving recorder has a fault and needs to be repaired in time.

[0043] In this embodiment, a ratio multiple mapping set between the number of shutter speed adjustments and the ratio reduction multiple is constructed, and the real-time number of shutter speed adjustments is input into the ratio multiple mapping set to obtain the ratio reduction multiple; the ratio adjustment operation means performing a product operation on the ratio reduction multiple and the initial shutter speed adjustment ratio; by the voice cleaning prompt, the possibility of the camera being dirty is excluded, the misreport rate of the driving recorder fault warning is reduced, and the camera lens can be processed in time, thereby improving the quality of the real-time images of the driving recorder.

[0044] Further, the specific content of image shutter adjustment is as follows: perform image blur analysis and classification on image fragments to obtain clear fragments and blurred fragments, and perform an influence ratio operation on the counted number of clear fragments and the number of blurred fragments to obtain an image adjustment ratio; compare the image blur influence index with the image influence threshold obtained from a preset database: if the image blur influence index is less than the image influence threshold, input the image adjustment ratio into the shutter adjustment mapping set to obtain a shutter speed adjustment multiple, and perform a shutter adjustment operation on the shutter speed with the shutter speed adjustment multiple to obtain a third shutter speed. The shutter adjustment mapping set represents the mapping relationship between the image adjustment ratio and the preset shutter adjustment multiple; if the image blur influence index is not less than the image influence threshold, input the blur influence deviation input ratio obtained by performing a deviation operation on the image blur influence index and the image influence threshold into the ratio mapping set to obtain a ratio adjustment value, input the adjusted image adjustment ratio obtained by performing an image adjustment ratio operation on the image adjustment ratio according to the ratio adjustment value into the shutter adjustment mapping set, obtain the corresponding shutter speed adjustment multiple to adjust the shutter speed to obtain a third shutter speed. The ratio mapping set represents the mapping relationship between the blur influence deviation and the ratio adjustment value.

[0045] In this embodiment, the influence ratio operation means performing a ratio operation on the number of clear fragments and the sum of the number of clear fragments and the number of blurred fragments. The shutter adjustment operation means performing a multiplication operation on the shutter speed adjustment multiple and the shutter speed. The image adjustment ratio operation means performing a multiplication operation on the ratio adjustment value and the image adjustment ratio. By fragmenting the locally blurred image, the accuracy of image blur analysis is improved. At the same time, through the dynamic regulation of the adjustment ratio, the shutter speed can be adjusted more real-time and accurately, thereby improving the quality of the real-time image of the driving recorder.

[0046] Specifically, the image influence threshold is obtained from a preset database. In a specific embodiment, substitute the initial judgment influence parameter corresponding to the clear image in the historical data into the specific limit expression of the image blur influence index to obtain a corresponding data set, and perform a mean operation on the data set to obtain the image influence threshold.

[0047] Further, the acquisition method of the image blur influence index is as follows: obtain the initial judgment influence parameter in real time and perform a unitless processing on the initial judgment influence parameter; perform a difference degree operation on the initial judgment influence parameter and the corresponding initial judgment threshold to obtain a difference value; obtain the image shutter influence weight from a preset database and perform a weighting operation with the image blur influence index to obtain an image blur influence degree value; perform an equilibrium operation on the total number of data types of the image blur influence degree value and the initial judgment influence parameter to obtain the image blur influence index. The image blur influence index is used to quantify the distance degree between the initial judgment influence parameter and the initial judgment threshold.

[0048] The specific limit expression of the image blur influence index is as follows: ; In the formula, i represents the data type number of the initial judgment influence parameter, , represents the i-th type of initial judgment influence parameter, represents the i-th type of initial judgment threshold, represents the image shutter influence weight corresponding to the i-th type of initial judgment influence parameter, represents the image blur influence index.

[0049] In this embodiment, the algorithm combines the initial judgment influence parameter, the initial judgment threshold, and the image shutter influence weight for comprehensive analysis to obtain the image blur influence index. In the formula, when the initial judgment influence parameter is smaller than the corresponding initial judgment threshold, it indicates that the influence degree of the initial judgment influence parameter on the imaging quality of the dash cam image based on the current shutter speed is lower, so the corresponding image blur influence index is smaller; the image blur influence index comprehensively quantifies the influence degree of the initial judgment influence parameter on the imaging quality of the dash cam, so as to perform corresponding image shutter adjustment in real time, and then obtain a dash cam image with higher image quality.

[0050] It should be added that when , the corresponding initial judgment influence parameter is the light intensity. When the light intensity is greater, the corresponding shutter speed is slower, which may cause overexposure or ghosting, affecting the image clarity and resulting in image blur; when , the corresponding initial judgment influence parameter is the vehicle speed. When the vehicle speed is faster, the corresponding shutter speed is slower, which may cause the vehicle's motion trajectory captured by the dash cam during exposure, resulting in image blur; when , the corresponding initial judgment influence parameter is the number of pedestrians. When the number of pedestrians is higher, the corresponding shutter speed is slower, which may cause the dash cam to capture the afterimage of pedestrians, resulting in image blur; when , the corresponding initial judgment influence parameter is the humidity. When the humidity of the vehicle driving environment is higher, the visibility of the vehicle is lower, resulting in image blur.

[0051] Specifically, the image shutter influence weight is obtained from a preset database. The image shutter influence weight represents the influence degree of the initial judgment influence parameter on the image blur influence index. Each initial judgment influence parameter and the image shutter influence weight have a unique mapping relationship, and the value range is between 0 and 1; for example, a mapping set of the initial judgment influence parameter and the preset image shutter influence weight is constructed, and the real-time initial judgment influence parameter is input into the mapping set to obtain the corresponding image shutter influence weight, representing the influence degree of the initial judgment influence parameter on the image blur influence index, and the sum of all image shutter influence weights is 1.

[0052] Further, when performing image blur analysis, if the trigger shutter impact adjustment is judged, the trigger shutter adjustment is performed, and the specific content is as follows: Input the counted shutter speed adjustment times into the real-time adjustment mapping set to obtain the corresponding shutter adjustment impact value. The real-time adjustment mapping set represents a set of mapping relationships between the counted shutter speed adjustment times and the preset shutter adjustment impact values; during the process of performing image blur analysis, if the trigger shutter impact adjustment is judged, and there is only one trigger judgment impact parameter, perform an initial adjustment impact operation on the initial deviation according to the shutter adjustment impact value to obtain the adjusted initial deviation; if there is more than one trigger judgment impact parameter, perform a coefficient adjustment operation on the shutter speed adjustment coefficient according to the shutter adjustment impact value to obtain the adjusted shutter speed adjustment coefficient.

[0053] In this embodiment, the initial adjustment impact operation represents performing a multiplication operation on the shutter adjustment impact value and the initial deviation, and the coefficient adjustment operation represents performing a multiplication operation on the shutter adjustment impact value and the shutter speed adjustment coefficient; by still performing the trigger shutter impact adjustment judgment during the process of performing image blur analysis, not only the timeliness of the shutter speed adjustment is ensured, but also the adjustment of the shutter speed is affected by the current analysis progress, further promoting the accurate adjustment of the shutter speed, thereby obtaining a clearer image of the driving recorder.

[0054] Further, during the process of performing image blur analysis, it also includes shutter adjustment feasibility analysis, and the specific content is as follows: When performing the trigger shutter impact adjustment judgment for the first time, obtain the preset shutter adjustment analysis time period. If within the shutter adjustment analysis time period, the number of times of performing the trigger shutter impact adjustment judgment reaches the preset judgment times, perform image capture based on the first shutter speed obtained from the last trigger shutter impact adjustment judgment and do not perform the trigger shutter adjustment judgment and image blur analysis until the shutter adjustment analysis time period ends.

[0055] In this embodiment, when the light intensity or vehicle movement changes violently (such as lightning, bumpy road surface), by pausing the image blur analysis, frequent adjustment of the shutter speed is avoided, the stability of the driving recorder work is improved, and at the same time, the storage pressure of the driving recorder is reduced.

[0056] The embodiment of the present application provides a real-time image analysis system for a driving recorder based on artificial intelligence, including: an initial analysis module, an image blur analysis module, and a trigger shutter adjustment module; wherein, the initial analysis module is used to perform shutter impact adjustment judgment based on the obtained initial judgment impact parameters, and adjust the shutter speed based on the judgment result; the image blur analysis module is used to perform image blur analysis on the driving recorder image to obtain an image quality analysis result, and take shutter speed adjustment measures according to the image quality analysis result; the trigger shutter adjustment module is used to perform trigger shutter adjustment if the trigger shutter impact adjustment judgment is triggered during the image blur analysis.

[0057] In this embodiment, through the quick pre-judgment of the initial analysis module and the precise repair of the image blur analysis module, the driving recorder realizes the leap from "passive recording" to "intelligent optimization", especially outstanding in complex lighting, weather and motion scenes. Its hierarchical processing architecture takes into account real-time performance, energy efficiency and imaging quality, not only provides a reliable technical basis for intelligent driving and traffic safety management, but also improves the imaging quality of the driving recorder and the analysis stability of the real-time image of the driving recorder.

[0058] In summary, the embodiment of the present application performs shutter impact adjustment judgment based on the obtained initial judgment impact parameters, adjusts the shutter speed based on the judgment result, then performs image blur analysis on the driving recorder image to obtain an image quality analysis result, and finally takes shutter speed adjustment measures according to the image quality analysis result. At the same time, if the trigger shutter impact adjustment judgment is triggered during the image blur analysis, trigger shutter adjustment is performed, so as to adjust the shutter speed more accurately, thereby improving the analysis stability of the real-time image of the driving recorder, and effectively solving the problem that the difference in driving states is not fully considered in the prior art during the process of the driving recorder taking pictures.

[0059] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0060] The present invention is described with reference to flowchart illustrations and / or block diagram illustrations of methods, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each flow and / or block in the flowchart illustrations and / or block diagram illustrations, and combinations of flows and / or blocks in the flowchart illustrations and / or block diagram illustrations, can be realized by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions executed by the processor of the computer or other programmable data processing apparatus create means for implementing the functions specified in the flow Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0061] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means that implement the functions specified in the flow Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0062] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in the flow Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0063] Although the preferred embodiments of the present invention have been described, additional changes and modifications can be made by those skilled in the art once they learn of the basic inventive concept. Therefore, the appended claims are intended to be construed to include the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0064] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A method for real-time image analysis of a driving recorder based on artificial intelligence, characterized in that, It includes the following steps: Based on the obtained initial judgment influence parameters, perform shutter influence adjustment judgment, and adjust the shutter speed based on the judgment result; Perform image blur analysis on the driving recorder image to obtain an image quality analysis result, and take shutter speed adjustment measures according to the image quality analysis result; When performing image blur analysis, if the shutter influence adjustment judgment is triggered, then trigger shutter adjustment is performed.

2. The real-time image analysis method of a driving recorder based on artificial intelligence according to claim 1, characterized in that: The process of performing shutter influence adjustment judgment based on the obtained initial judgment influence parameters and adjusting the shutter speed based on the judgment result is as follows: Perform a difference operation on the initial judgment influence parameters and the corresponding initial judgment thresholds obtained from the preset database, and record the initial judgment influence parameters corresponding to the difference operation result not less than 0 as the trigger judgment influence parameters; If there is only one trigger judgment influence parameter, input the initial deviation obtained based on the trigger judgment influence parameter and the corresponding initial judgment threshold into the first shutter adjustment mapping set to obtain the corresponding first shutter adjustment multiple, and perform a shutter speed adjustment operation based on the first shutter adjustment multiple and the shutter speed to obtain the first shutter speed. The first shutter adjustment mapping set represents the mapping relationship between the initial deviation and the preset first shutter adjustment multiple; If there is more than one trigger judgment influence parameter, quantify and evaluate the influence degree of the shutter speed on the image imaging based on the initial deviation between the trigger judgment influence parameters and the corresponding initial judgment thresholds to obtain a shutter speed adjustment coefficient, input the real-time shutter speed adjustment coefficient into the second shutter adjustment mapping set to obtain the corresponding second shutter adjustment multiple, and perform a shutter speed adjustment operation on the shutter speed based on the second shutter adjustment multiple to obtain the first shutter speed. The second shutter adjustment mapping set represents the mapping relationship between the shutter speed adjustment coefficient and the preset second shutter adjustment multiple; If there are no trigger judgment influence parameters, perform image blur analysis.

3. The real-time image analysis method of a driving recorder based on artificial intelligence according to claim 2, characterized in that: The specific process of obtaining the shutter speed adjustment coefficient is as follows: Number the obtained trigger judgment influence parameters according to the data type; Perform an initial deviation operation on the trigger judgment influence parameter and its initial judgment threshold to obtain the corresponding initial deviation, and perform an influence weighting operation on the shutter speed adjustment weight obtained from the preset database and the obtained initial deviation to obtain an initial deviation influence value; Perform an equalization operation on the initial deviation influence value to obtain a shutter speed adjustment coefficient, and the shutter speed adjustment coefficient is used to quantify the influence degree of the shutter speed on the image imaging.

4. The real-time image analysis method of a driving recorder based on artificial intelligence according to claim 1, wherein: The process of performing image blur analysis on the driving recorder image to obtain an image quality analysis result and taking corresponding shutter speed adjustment measures according to the image quality analysis result is as follows: Perform image blur analysis on the real-time image of the driving recorder to obtain an image quality analysis result, and the image quality analysis result includes global blur, local blur, and clear images; If the image quality analysis result is global blur, adjust the shutter speed according to the initial shutter adjustment ratio to obtain the second shutter speed. If the image quality analysis result obtained based on the second shutter speed is global blur, take shutter iteration adjustment measures; If the result of the image quality analysis is local blurring, the image blurring effect is evaluated based on the initial judgment influence parameters to obtain an image blurring influence index. The real-time image blurring influence index is input into the image segmentation mapping set to obtain the corresponding number of image segmentations. Image segmentation is performed according to the number of image segmentations to obtain image fragments, and the image shutter is adjusted. The image segmentation mapping set represents the mapping relationship between the image blurring influence index and the preset number of image segmentations; If the result of the image quality analysis is a clear image, no additional processing is performed.

5. The real-time image analysis method of a driving recorder based on artificial intelligence according to claim 4, characterized in that: The specific content of the shutter iterative adjustment measure is as follows: If the result of the image quality analysis based on the second shutter speed is global blurring, the reduction multiple is mapped according to the number of shutter speed adjustments. The adjustment ratio is obtained through the proportional adjustment operation of the reduction multiple and the initial shutter adjustment ratio, and the second shutter speed is adjusted according to the adjustment ratio; When the number of adjustments of the second shutter speed reaches the preset number and the result of the image quality analysis is still global blurring, a voice cleaning prompt is given; When the number of adjustments of the second shutter speed reaches the preset number and the result of the image quality analysis is local blurring, local blurring analysis is performed; When the number of adjustments of the second shutter speed reaches the preset number and the result of the image quality analysis is a clear image, no additional processing is performed; After the voice cleaning prompt, the cleaning effectiveness judgment is also included. The specific process is as follows: After the voice cleaning prompt, if the result of the image quality analysis is global blurring, a fault warning is given. If the result of the image quality analysis is local blurring, local blurring analysis is performed. If the result of the image quality analysis is a clear image, no additional processing is performed.

6. The real-time image analysis method of a driving recorder based on artificial intelligence according to claim 4, wherein: The specific content of the image shutter adjustment is as follows: Image blurring analysis and classification are performed on the image fragments to obtain clear fragments and blurry fragments, and the influence ratio operation is performed on the counted number of clear fragments and the number of blurry fragments to obtain an image adjustment ratio; If the image blurring influence index is less than the image influence threshold, the image adjustment ratio is input into the shutter adjustment mapping set to obtain a shutter speed adjustment multiple. The shutter speed is adjusted through the shutter speed adjustment multiple to obtain a third shutter speed. The shutter adjustment mapping set represents the mapping relationship between the image adjustment ratio and the preset shutter speed adjustment multiple; If the image blurring influence index is not less than the image influence threshold, the blurring influence deviation obtained by performing a deviation operation on the image blurring influence index and the image influence threshold is input into the ratio mapping set to obtain a ratio adjustment value. The adjusted image adjustment ratio obtained by performing an image adjustment ratio operation on the image adjustment ratio according to the ratio adjustment value is input into the shutter adjustment mapping set to obtain the corresponding shutter speed adjustment multiple to adjust the shutter speed to obtain a third shutter speed. The ratio mapping set represents the mapping relationship between the blurring influence deviation and the ratio adjustment value.

7. The real-time image analysis method of a driving recorder based on artificial intelligence according to claim 4, wherein: The acquisition method of the image blurring influence index is as follows: The initial judgment influence parameters are obtained in real time, and the unit of the initial judgment influence parameters is removed; The difference value is obtained by performing a difference degree operation on the initial judgment influence parameters and the corresponding initial judgment threshold; Obtain the image shutter impact weight from a preset database, and perform a weighted operation with the image blur impact index to obtain the image blur impact degree value; Perform an equalization operation on the total number of data types of the image blur impact degree value and the initial judgment impact parameter to obtain the image blur impact index, and the image blur impact index is used to quantify the distance degree between the initial judgment impact parameter and the initial judgment threshold.

8. The real-time image analysis method of a driving recorder based on artificial intelligence according to claim 1, characterized in that: When performing image blur analysis, if the shutter impact adjustment judgment is triggered, trigger shutter adjustment, and the specific content is as follows: Input the counted number of shutter speed adjustments into the real-time adjustment mapping set to obtain the corresponding shutter adjustment impact value, and the real-time adjustment mapping set represents a set of mapping relationships between the counted number of shutter speed adjustments and the preset shutter adjustment impact value; During the process of performing image blur analysis, if the shutter impact adjustment judgment is triggered and there is only one trigger judgment impact parameter, perform an initial adjustment impact operation on the initial deviation according to the shutter adjustment impact value to obtain the adjusted initial deviation; If there is more than one trigger judgment impact parameter, perform a coefficient adjustment operation on the shutter speed adjustment coefficient according to the shutter adjustment impact value to obtain the adjusted shutter speed adjustment coefficient.

9. The real-time image analysis method of a driving recorder based on artificial intelligence according to claim 8, characterized in that: During the process of performing image blur analysis, it also includes shutter adjustment feasibility analysis, and the specific content is as follows: When performing the shutter impact adjustment judgment for the first time, obtain the preset shutter adjustment analysis time period. If the number of times of performing the shutter impact adjustment judgment reaches the preset judgment number within the shutter adjustment analysis time period, perform image capture based on the first shutter speed obtained from the last shutter impact adjustment judgment without performing shutter adjustment judgment and image blur analysis until the shutter adjustment analysis time period ends.

10. A real-time image analysis system for a driving recorder based on artificial intelligence, characterized in that, Including: Initial analysis module, image blur analysis module, and trigger shutter adjustment module; Among them, the initial analysis module is used to perform shutter impact adjustment judgment based on the obtained initial judgment impact parameter and adjust the shutter speed based on the judgment result; The image blur analysis module is used to perform image blur analysis on the dashcam image to obtain an image quality analysis result, and take shutter speed adjustment measures according to the image quality analysis result; The trigger shutter adjustment module is used to perform trigger shutter adjustment if the shutter impact adjustment judgment is triggered during image blur analysis.

Images