Affine invariant features image matching approach based on principal components analysis

A comparison of the Scale-Invariant Features Transform approach and Harris-Affine Invariant features approach was studied theoretically and experimentally. To improve the speed of image matching and accuracy of image features descriptors, an approach of reducing the dimensions of image matching features descriptors was proposed, which was applied to Harris-Affine region detectors based on Principal Components Analysis. Experimental results show that the method can accelerate the Harris-Affine Invariant features approach. On the premise of not degrading the matching effect, the method can reduce the dimensions of features vectors from 128 to 15.