Study On Front Vehicle Detection From Video Based On Motion Vector Analysis

Recently,auto-driving has been paid more and more attention with the rapid development of intelligent transportation,in which front vehicle detection is one of the important components.It can avoid people’s subjective consciousness and behavior.What’s more,it can also predict the traffic congestion ahead of the situation,planning a smooth and convenient travelling path.Therefore,it is of great significance to carry out the research work of the front vehicle detection.In this paper,according to the characteristics of video-image,the research work on the front vehicle detection is carried out by the following aspects:First of all,we introduce the background and the current situation of the intelligent transportation,and briefly describe the significance of the research.Secondly,several methods of vehicle detection both in static and dynamic scenes have been introduced.We focus on the analysis of the moving vehicle detection method in dynamic scenes,combining with the characteristics of the video-image,we analyze the improvement ideas of the current algorithms and introduce the concept of motion vector.Thirdly,in order to obtain the motion vector,a method based on spatial constraint has been proposed.Combining with the characteristics of video-image,this paper analyzes the shortcomings of the brute-Force matching algorithm and assigns a certain weight to spatial position,the similarity is represented by the gray levels and spatial location.Finally,inspired by the transmission relationship between spatial coordinate and image plane,a front vehicle detection method based on motion vector and vanishing point has been proposed,whose good accuracy and robustness have been showed by a large number of tests.