Abstract: In order to improve the rapidity of object detection with circular outlines,an improved symmetry-based Hough transform algorithm for circle detection is proposed. First, an input image is denoised by Gaussian algorithm, and its edge is detected with Canny algorithm. At the same time, the possible collinear points are removed. Then, based on circle symmetry, the coordinate of circle center is got by using Hough transform in one-dimensional space. Finally, voting with Hough transform is employed to obtain radius in one-dimensional space. Comparative experiment results indicate that the proposed algorithm has a higher recognition accuracy and quicker speed of operation.