Investigation On Iterative Closest Points Method Based On Robust Statistic Method

The visual system is one of the key subsystems of the IC manufacturing equipment.Image matching is function module of core in visual system for IC manufacturingequipment. Iterative closest point algorithm is a matching method for point set, which hasa wide range of applications because of its high registration accuracy in image registration.This article study how to improve the iterative closest point algorithm in two stages of theweight given to the point pairs and the elimination of specific point pairs.First, the article describes the basic principles of the iterative closest point algorithmand analyzes the characteristics and shortcomings of the iterative closest point algorithm.In the two stages of the weight given to the point pairs and the elimination of specificpoint pairs, the robust statistical methods can improve the robustness of the iterativeclosest point algorithm. The main purpose of Robust Estimation is that robust estimation isable to make a good estimate when there are outliers in the observational data. The paperintroduces three classic iterative closest point algorithms that incorporated the robuststatistics methods, i.e., M-ICP algorithm, the RICP algorithm and TrICP algorithm.Another method of LTS is combined with the iterative closest point algorithm isproposed, which utilizes residuals distribution characteristic to get the number of pairs ofthe specific point selected in each iteration and calculation of the scaling parameter isjoined. Experimental results show that the new algorithm for point set matching is feasibleand effective in the presence of noise.