Design And Implementation Of Reversing Radar Image System Based On Information Fusion

With the progress of society and the improvement of people's living standards,cars are becoming more and more popular.While providing convenience to people,it has also caused a series of traffic safety accidents,threatening people's lives and property.Arouse people's attention.Through research on sensor technology and reversing assist technology at home and abroad,this paper designs an information fusion reversing radar imaging system around reversing safety issues.It not only makes full use of the advantages of different sensors,but also fuses the sensor information to make the decision information of reversing more accurate and reliable.The main contents of this paper are as follows:First,according to the actual application scenarios of the system and specific needs analysis,the overall design of the system was carried out,including,the collection of sensor data,data fusion processing and the output of final information.Secondly,use binocular vision and ultrasonic reversing radar to build a distance sensing platform for reversing environment.For binocular vision,two wide-angle cameras are used to collect the image of the reversing environment.First use the distortion correction method to correct the wide-angle image,so that the corrected image can retain a lot of edge information.Then,the stereoscopic correction and stereo matching are used to obtain the parallax map of the reversing environment.Obtain the depth information of the obstacle using the principle of triangulation after the position of the obstacle.At the same time,a correction method is given for the accuracy problem of binocular ranging.For the ultrasonic reversing radar,for the single-receiving single-issue reversing radar,the measurement is located between the two probes For the problem of large distance measurement results when obstacles are present,an improved method for triangular distance measurement is given.Thirdly,the information fusion algorithm of the sensor is designed.According to the different ranging range of the ultrasonic radar probe,the spatial alignment method of the radar probe and the binocular vision sensor is designed.By studying the different characteristics of the two sensors,the two sensors can complement each other's advantages,and a data-level fusion algorithm based on two-stage Kalman filtering is designed.The first-stage filter is used to reduce noise interference,and the second-stage filter obtains the optimal result through mutual correction of the two sensor data.Finally,the software and hardware platform was built for the system operation,the embedded Linux operating system was selected on the Smart210 development board to the OpenCV visual library transplanted as the application dependent library,the binocular video acquisition program was design based on V4L2,The programming method realizes the collection of two kinds of sensor information,and realizes the transplantation of information fusion algorithm.At the same time,in an experimental environment with good light and no visual occlusion,the system performance is tested and verified with a variety of obstacles under the movement of about 1.5m/s speed.The test results show that the system reaches the predesigned performance indicators and can meet the reliability and real-time requirements in the reverse environment.