| As the continuous improvement of living standards, automobile has become an indispensable transport tool in the daily life. Since the21th century car ownership growth rapid in the world. The management of vehicle information and road transport facilities facing immense challenges. So it has entered a new era of intelligent transportation system. Vehicle logo is an important part of vehicle information, Vehicle logo recognition system is the significance of vehicle information management.Vehicle logo recognition system includes image capture, image preprocessing, vehicle logo detecting and recognition. Vehicle logo detecting and recognition plays a key role, therefore, this paper study the detecting and recognition of the vehicle method. The paper major work as follows:(1) Analysis vehicle logo color characteristics, Gray space is appropriate for vehicle logo. Then denoising and normalization.(2) Vehicle logo detecting, Firstly, Use gradient suppress horizontal or vertical texture around the logo, the maxima minimum filtering can highlight the logo area, then, horizontal and vertical project gradient image, detecting vehicle logo, finally adopted symmetry verify the logo area. The experimental verify this method has better logo detecting Result.(3) Vehicle logo recognition, Firstly, Achieve the2DPCA to reduce the image dimension and using KNN for recognition, Secondly, proposed extraction HOG features of vehicle log image and using KNN/SVM for recognition experiments. The experiments result show HOG with SVM are better than others, HOG features is a better description of the vehicle logo shape.(4) System design. For the above algorithm to design a vehicle logo recognition system and select DM8168DaVinci digital media processor as the algorithm running platform, now have completed part of the algorithm transplantation. |
PDF Full Text Request