Image object recognition based on Radon transform

Abstract

The processing of image object of visual sensor is a key issue in robot intelligent control. A new method for object recognition based on Radon transform is proposed. On the basis of image preprocessing, the algorithm for extracting the shape features of image object at different angles based on Radon Transform (RT) is given. To increase the efficiency of the algorithm, two feature sets have been formed. One is coarse feature set that is formed to measure the similarity between the template object and image object and then the template and image are aligned. Another is fine feature set that is formed to realize the object recognition by feature matching. The results of some tests show that the new method is robust to the translation, rotation and scale of image object.