关键词: 滑坡形变场, 高分影像, 多光谱融合影像, SIFT算法, CSIFT算法

Abstract: The feature points of panchromatic remote sensing images can be extracted by using SIFT (scale-invariant feature transform) algorithm，but the feature points is characterized by small quantity and uneven distribution，besides that，the monitoring effects of landslide deformation field is influenced to some extent.In order to improve the monitoring precise of landslide deformation field，a new method of calibrating landslide deformation field based on SIFT algorithm and CSIFT (colored scale invariant feature transform) algorithm is proposed.According to the principle of the new method，the panchromatic image and multispectral fusion image are processed by using SIFT algorithm respectively，the multispectral fusion image is processed by using CSIFT algorithm，the different feature points extracted by the above processes are superimposed，so as to obtain more feature points.Taking the slope of the south side of Fushun West open-pit coal mine as an example，the test results show that the new method can fully utilize the characteristics of the three different image matching processes，the quantity of feature points are increased greatly，which make the generated landslide deformation field more detailed and accurate，and the extent of landslide more accurate.The study result further show that the new method is characterized by higher degree of automation and lower cost，therefore，it is very suitable for monitoring large deformation field of large-scale landslide.

Key words: Landslide deformation field, High-resolution image, Multispectral fusion image, SIFT algorithm, CSIFT algorithm