Studying On Vehicle Detection And Recognition Method For Monocular Camera Vision System

With the gradual improvement of intelligent unmanned vehicle systems functions, the real-time requirements for processing road information are getting higher and higher. Recognition on vehicle type will expand unmanned vehicle systems’ application in the actual and complex environment. In addition, vehicle type identification plays a very important role in the development of target recognition field and intelligent transportation system. However, there are less researches on vehicle type recognition in the areas of target recognition and intelligent transportation systems.In this paper, image sequences consisting of vehicles acquired by vehicle monocular camera are chosen as an object library. The following work and research are focused on the two tasks about vehicle detection and identification for classification.1. Proposed an algorithm which is a fast vehicle detection algorithm based on the channel features and cascaded Ada Boost classifier. Then we applied it in detecting an actual vehicle combined with a sliding window.2. Proposed an algorithm which is an efficient vehicle recognition algorithm based on Dictionary Learning algorithm using Histogram Intersection Kernel. Firstly, we used this algorithm to get the effective description code of local feature. Then we use the least squares probabilistic classifier to achieve the rapid identification of multi-kinds of vehicle types.All the above work and achievements for the research of vehicle detection and recognition are based on intelligent vehicle driving. On one hand, the study has made an active exploration in the practical application and the final results confirmed the effectiveness of this method. On the other hand, this research has also established a foundation for the studying of the real-time target recognition in a complex driving environment.