Improved Randomized Hough Transform Based on Circular Power Theory

Circle extraction is a fundamental task in computer vision, which is widely applied in automated inspection and assembly. However, existing circle detection algorithms often exhibit limitations in terms of noise resistance and computation speed. This paper addresses these challenges by proposing a novel approach to circle detection, which is based on a clustering and fitting method in the context of computer vision assignments. The main objective of this research is to enhance the performance of the Randomized Hough Transform, a widely used technique for circle detection. To achieve this, the proposed method leverages the concept of circular power theory to identify potential circles and mitigate the accumulation of invalid cells associated with multiple circles. By selecting circles in which three points align on the same circle, the proposed method effectively reduces the accumulation of invalid cells during the random sampling process. Experimental results demonstrate the improved robustness and computational accuracy of the proposed circle detection method on both synthetic data and real-world images.