A New Method of Speeded Up Robust Features Image Registration Based on Image Preprocessing

摘要

Target recognition is a challenge and one of the most active research areas in computer vision. Variations in image cause many challenges to target recognition system due to change in illumination, noise and detail blurring. In this paper, a new approach based on image preprocessing is proposed to speeded up robust features in target recognition. The proposed image preprocessing method improves the global air-light first by using Retinex algorithm of the original image, then a homomorphic filtering scheme is proposed after db2-type wavelet transform. In target recognition, based on speeded up robust feature extraction approach, the feature points are extracted by Hessian operator in the algorithm, then the algorithm gets all the matching pairs by matching threshold. The experiments show that the algorithm can solve the registration accuracy problem under complex conditions, and improve the correction results effectively.