Application of Hough Transform in lateral multi-lens video logging image segmentation

摘要

Hough Transform is a well-known effective method for circle detection. However, Hough Transform requires massive storage space and huge cost in computation time. For the purpose of real-time image segmentation during lateral multi-lens video logging, this paper presents an algorithm for quick circle detection based on Hough Transform. The proposed method is consisted of three sequential steps. First, a chain code of the targeting image is computed by boundary detection method. Secondly, a candidate circle is determined by picking three points from the chain code following a certain rule. Finally, the candidate circle is identified by referring to the available radius. The following experiments show the proposed algorithm can effectively solve the issues of Hough Transform, i.e. small storage space and less computation time. What is more, more than one circle could be identified in the targeting image.