Spot Position Extraction Based on High-resolution Parametric Subspace Method without Eigen decomposition

Abstract

This paper presents a subspace method of spot centroiding algorithm for locating the centers of laser spots. It focuses on how to find the position of the activated pixel which is the position of the imaged spot on the detector of the camera using Gaussian Approximation method (GAM), Center of Mass (COM) method and Blais and Rioux algorithm. The traditional subspace methods based on either singular value decomposition (SVD) or eigen decomposition (ED) to estimate the signal or noise subspace, these decomposition methods tend to be computationally intensive. In order to improve the algorithm of spot position determination, a proposed subspace analysis is used to determine spot position without eigen decomposition. Therefore, a method of fast subspace decomposition using Lanczos algorithm is presented and compared with other methods. A simulated example is provided to evaluate the proposed method and the simulation results shows that the proposed algorithm has the advantages of reduction the computational load and superior estimation performance.