A New NAM-based Hough Transform Algorithm

This paper proposes an improved Hough Transform algorithm based on the non-symmetry and anti-packing pattern representation model (NAM). It is a kind of Hough Transform based on blocks, which means that the smallest voting unit is a block but not a pixel point. Obviously, the smaller the number of blocks is, the more efficient the algorithm is. Because NAM usually has a smaller number of blocks divided than that of other dividing methods such as binary tree, quadtree and IBR (image binary representation), NAM can significantly improve the efficiency of Hough Transform. Besides, in this paper, NAM is combined with run-length smearing algorithm (RLSA)to reduce noise and further reduce blocks. In experiments, this algorithm is compared with IBR-based Hough Transform. The result proves that NAM-based Hough Transform with RLSA is more efficient and accurate.