A Real-Time Ellipse Detection Method Using GPU-Based RANSAC

An approach to very rapid and real-time ellipse detection is described, Based on a modified RANSAC (RANdom Sample Consensus), and considering the parallel processing capabilities of a programmable graphics processing unit. The method is able to detect ellipses of images in real-time. First, it finished the computational part of image segmentation and contour detection on the CPU. Then it uses the fragment processing capability of the GPU to fit ellipse shape parameters. Experimental results show that the computational efficiency of the proposed method will be multiplied several times than other algorithms, and this method is more efficient for real-time ellipse detection.