Circle detection of short arc based on Randomized Hough Transform

Abstract

Detecting circle based on short irregularly arc in complex background and illumination condition has a wide industrial application, which provides the theoretical basis for some special intelligent instrument. Aiming at the problems of small uneven arc, uneven illumination, light intensity variation, blur boundary and adjacent grey degree, this paper proposes a circle detection method based on Randomized Hough Transform (RHT) combined with the circle features and gradient algorithm, which can accurately detect the edge image, locate the arc position and detect the circle under complex condition. The steel pipe images in the storage field were detected and analyzed. The results prove that the proposed algorithm can accurately detect the edge image, locate the arc and detect the circle under complex condition. Compared with the existing commonly used detection algorithms, this method can achieve higher recognition effect and fully satisfy the requirements of practical industrial applications.