| Video monitoring has been widely used in the field of industry, military, image processing and the intelligent monitoring, etc. With the development of human society, science, economy and the Internet, people's demands increase continuously. To meet the accuracy requirements, the specific targets recognition is especially becoming more and more important. This paper mainly researches on the detection of specific vehicle targets, distinguishing the specific vehicles among other moving targets, further adapted to the needs of the reality. To achieve the purpose of the specific targets recognition, two questions must be solved: the moving targets detection and the moving targets recognition.The moving targets recognition is based on the moving targets detection, this paper proposes an improved background difference arithmetic after an in depth research and analysis. This new arithmetic increases dynamic threshold, optimizing the selection of mean value about background and making the front targets more accurate and clear. The experimental results show that the improved algorithm has the higher precision, and also insures instantaneity.This paper mainly studies two kinds of methods of the targets classification: Shape information classification method and Movement characteristics classification method. After comparing the two kinds of methods, this paper selects the shape information classification method and finally chooses color, aspect ratio to distinguish the specific vehicles among other moving targets. Then, the detection target values are extracted and matched to the values in standard images. This paper proposes a characteristic parameter include the magnitude and phase of Zernike moment, realized the specific vehicle identification. In addition, experiment is carried out to verify.Finally, this paper summarizes the shortcomings and insufficiencies in the current targets detection and specific targets recognition, at the same time, presents the future development. |
PDF Full Text Request