GA-SURF: A new Speeded-Up robust feature extraction algorithm for multispectral images based on geometric algebra

Abstract

Speeded-Up Robust Features (SURF) can be used for tasks such as object recognition, image registration, classification or 3D reconstruction. However, existing SURF algorithm cannot be directly applied to deal with multispectral images. In this paper, based on SURF and the theory of Geometric Algebra (GA), a novel feature extraction algorithm named GA-SURF is proposed for multispectral images. First, a Hessian matrix based on GA is calculated for locating interest points in spatial and spectral space. The box filters based on GA are used to simplify the calculation of Hessian matrix and generate image pyramids. Then, following the procedures of SURF, interest points are located by the image pyramids and described in GA space. Experimental results demonstrate that, compared with other feature extraction algorithms for multispectral images, GA-SURF can be computed much faster and are more robust and distinctive.