Research Of Front Vehicle Detecting And Recognition Technology Based On Intelligent Pure Electric Vehicle

With the acceleration of urbanization process and the development of people's living standards,the vehicle population of China is also increasing.Meanwhile,the problems of Urban traffic congestion,traffic accidents and environment pollution are increasing.Intelligence could help with driving safety and convenience,while electrification helps a lot on vehicle emission pollution.The fusion of Intelligence and electrification has been an important trend of vehicle development.Detection method of front vehicle is the key technology of vehicle intelligence by improving driving safety,which is worthy being researched.Detecting methods differs from vehicle to vehicle due to volume and features.Thus,saloon car was chosen to be detected object.First of all,this article gathered and preprocessed the data of Laser Radar,describing the principle of clustering algorithm in detail.The image gathered by Industrial Camera was preprocessed as well,including graying,denoising and division.Multi-sensor data fusion model detecting front saloon has been established,mixing those two sensor data in space and time.The calibration of Laser Radar and Industrial Camera have been finished,realizing data conversion under different coordinate systems.Second,the article proposed a method of front saloon detecting based on the fusion of Laser Radar and Industrial Camera,using adaptive distance threshold clustering method,clustered Laser Radar data and got its interested region.Although laser radar can accurately detect the distance and position information of the front obstacles,it can't detect the characteristic of obstacles,so the other obstacles which have the similar size and appearance may also be detected.Then the laser radar region of interest should be projected into the image by coordinates conversion,generating the interested region of image.The article also proposed a method of image feature extraction based on hierarchical structure,extracting features of saloon such as shadow of saloon,contour symmetry and texture complexity.Then multiple feature fusion algorithm was proposed based on Mahalanobis distance method,detecting front vehicle by assigning different values into every feature.Last,this article established front saloon detecting platform and finished hardware installation including Laser Radar,Industrial Camera and IPC(Industrial Personal Computer).The system software procedure and software detecting platform were designed as well.Last,front saloon was detected under urban,freeway and raining driving cycles,and the accuracy was counted..According to the test,the rate of accuracy of saloon detecting could above 80%.