عنوان مقاله [English]
نویسندگان [English]چکیده [English]
In vehicle surveillance systems, one of the appropriate methods for recognition are 3-D models. Several methods have been proposed for this purpose. Feature based methods are most significant and widely used. In this paper, is proposed an algorithm within recognition framework. Proposed algorithm is considered information of image and model edges as feature. A block descriptor has been used extract edges information to feature vector. Every feature vectors provide arrangement and layout in neighbourhood of edge point. Image and model feature vectors are compared using nearest neighbour method and measuring compliance are stored in a score matrix. Finally, the model has the most points in the image is detected as vehicle type. The experimental result is shown the proposed algorithm in terms of speed and accuracy offers better performance than the algorithms SURF and FREAK.