Matching Low Altitude RS Image with Harris-Laplace and SIFT Descriptor

An improved feature matching method based on Harris-Laplace and SIFT descriptor is proposed.Because of the instability of low altitude remote sensing platform,the difference of spin deflection angle and scale between low altitude remote sensing imageries is great.The results obtained by matching method based on area grayscale can't meet the real requirements.The feature points detected by SIFT algorithm are easily affected by image noise and slight texture change.With the proposed method,Harris-Laplace is used to detect key points of the image,which are invariant to illumination change,image noise and scale change.And then the orientation of these key points is determined to form feature points from these key points.These feature points are described by SIFT descriptor,and matched using BBF algorithm and RANSAC algorithm.One experiment is introduced,which uses low altitude remote sensing image with high resolution as input data.The experimental results show that the proposed method possesses higher matching accuracy at the same matching speed compared with the matching method based on SIFT algorithm.