Research On Key Target Detection Algorithm Based On Radar And Camera Information Fusion

With the increasing demand of consumers for vehicle safety and comfort,intelligent driving technology has received extensive attention and research.As an important part of intelligent driving system,the environmental information acquired by the sensing module is the basis for the successful application of intelligent driving technology.The mainstream sensors of intelligent driving systems currently in mass production are radars and cameras.Radars have high accuracy in detecting target distance and speed and can work around the clock,but they cannot accurately identify the type of target and are easily affected by noise.The camera can obtain rich environmental information,and its price and technical difficulty are relatively low,but it is more sensitive to changes in light and weather.So,a single sensorbased environment sensing solution may often have some limitations.Through information fusion of different sensors,the target detection advantages of each sensor can be fully utilized to achieve information redundancy and complementarity.Therefore,in order to improve the target detection capability of the system,the research on multi-sensor information fusion is particularly important.Accordingly,the environment perception module of the intelligent driving system is the research object of this article.An information fusion scheme based on millimeter wave radar and camera is designed.The millimeter wave radar and camera are used for target detection,and then feature extraction,identity recognition and information fusion are performed on the target data.And then conduct risk assessment and screening of dangerous targets for the fusion target.The main research contents are as follows.(ⅰ).Develop a target detection process based on the fusion of millimeter wave radar and cameraAnalyze the advantages and disadvantages of different sensors in target detection,and the advantages and disadvantages of different levels of fusion methods,then analyze the target detection and information fusion schemes proposed by scholars,explore the research points that need to be improved,and finally develop a millimeter wave-based Radar and camera fusion sensor scheme.(ⅱ).Preprocessing sensor dataAccording to the similarity of data,the target information of MMW radar is clustered,and the multiple radar reflection points of the same object are divided into one cluster,so as to obtain more accurate target information.Then,for the problem of random noise in the original data of the sensor,the original data was filtered according to the motion characteristics of the target before information fusion,so as to improve the signal quality and reduce the influence of the interference signal on the subsequent algorithm.(ⅲ).Design the fusion algorithm architecture based on millimeter wave radar and cameraAiming at the inconsistency of the reference coordinate system used by each sensor when detecting the target,before performing fusion processing on the target data of different sensors,first perform time registration and spatial registration on each sensor to ensure different sensor data obtained by the fusion center describes environmental information at the same time and location.Then the target list of different sensors is matched according to certain rules,and the measurement data of each sensor on the same target is fused to obtain the complete target information.Finally,the fusion target is tracked to improve the accuracy and reliability of target detection.(ⅳ).Develop the dangerous target screening strategyTarget risk assessment is often related to the establishment of target motion models and target motion prediction.Therefore,first analyze the advantages and disadvantages of various target motion models and target motion prediction methods,and decide to use behavior-based motion models,and then use the historical trajectory of the vehicle to determine its driving status.Then calculate the collision time of the vehicle based on the relative distance and speed between the front vehicle and the front vehicle,so as to evaluate the danger of the fusion target,and then select the target with a higher risk of collision,and then provide reference information for the decision of the subsequent control system.(ⅴ).Experiment and result analysis of the algorithm under different working conditionsSet up an experimental platform for multi-sensor fusion and perform relevant tests on typical urban roads and urban expressways.Then select several of the more representative working conditions to analyze the test results,compare the detection results of radar targets,camera target detection results,and target detection results of fusion of radar and camera under different working conditions.At the same time,the rationality of the dangerous target selection strategy is verified.The results show that the algorithm proposed in this paper can effectively improve the stability and reliability of target recognition and reduce the probability of missed detection.