Fast algorithm of detecting circular edge based on multiresolution wavelet transform

In order to accelerate the speed of traditional Hough Transform of detecting and locating the circular object in an image, a new approach has been introduced in this paper which is based on multiresolution wavelet transform. By utilizing odd order B_spline wavelet function, the image dimension can be reduced; the gradient and tangent angle of the image pixels could be calculated; and the image edges can be extracted as well. Since tangent angles of an image have been calculated, the dimension of parameter space in Hough Transform also reduces. Therefore, after preprocessing the original image by means of B_spline wavelet Transform, we can detect the circular object fast and precisely through improved algorithm of improved Hough Transform. By analyzing the process of this algorithm, we can find that the computation of the algorithm can decrease in proportion to the exponential function of image scale. Finally, we demonstrate our method through experiment test which is realized by programming on matlab.