Space Targets Classification and Recognition Based on Affine Invariant Moments

Abstract

An affine transformation model is established for ground-based photoelectric imaging and detecting systems. This model shows that affine transform consists of five basic transformations, which are translation, scale, slant and rotation transformations. Six affine invariant moments are given with geometrical moments. Furthermore. affine invariant moment vectors of training data are calculated for four-class simulated space target images. For different viewpoint angle target images, the affine invariant moments are calculated. Simulation experiments are performed for space targets classification and recognition through their eigenvectors comparing. Simulated results show that the method of target images classification recognition based on affine invariant moments is effective. Keywords-Space targets, Affine transform model, Affine invariant moments, Classification, Recognition