Remote Sensing Image Registration Algorithm Based on Multi-scale Corner and Haar Transformation

[摘要] This paper proposed a new image registration algorithm based on multi-scale corner and Haar transformation, which its core idea is the combination of Haar operator and the Laplace image pyramid are used in down sampling the original images, and detect corner at the resulting images, Angle of the corner and most value point of scale response (scale precision positioning) will get, registrating images will obtained after rotation and translation. Theoretical analysis and simulation results shows the algorithm has higher registration accuracy and small amount of computation.

[发布机构] Xi'An Research Inst. of Hi-Tech, Shaanxi, China^1

[学科分类] 环境科学（综合）

[关键词] Analysis and simulation;Down sampling;Image registration algorithm;Laplace image;Original images;Precision positioning;Registration accuracy;Remote sensing images