Radioactive Waste Barrel Localization Based on Concentric Circle Detection with Edge Pre-Classification and Randomized Hough Transform

In this paper, we propose an effective method to locate radioactive waste barrel with imperfect concentric circle edges. First, this method estimate the rough centers of circles quickly based on lookup table and bitwise accumulator. Based on the rough centers and the gradient of edges, the candidate edges of circles are collected and pre-classified into overlapped detection domains. After the edge classification, the best centers of each domain are voted by Randomized Hough Transform. According to the votes of the best centers, the true center of concentric circles is determined by the centers with the highest concentricity. We experimentally compare the performance of the proposed method with RANSAC and LUT-BCA. The results show that proposed method is fast, accurate and robust.