Small UAV based multi-viewpoint image registration for monitoring cultivated land changes in mountainous terrain

Land degradation, soil erosion, and illegal occupation in the mountainous terrain of southern China have severely reduced the amount of cultivatable land. The use of small unmanned aerial vehicles (UAVs, aka drones) equipped with various types of cameras is considered to be a flexible and low-cost platform for monitoring cultivated land changes. However, image pairs of the same scene taken from different viewpoints often contain discontinuous rotated images with illuminated variations. To address these problems, a novel small UAV based multi-viewpoint image registration method for monitoring cultivated land changes in mountainous terrain is proposed. First, a mixed feature descriptor (MFD) is defined for measuring global and local discrepancies between two datapoint sets, and a deterministic annealing scheme is employed to control the balance of the MFD. Second, the mixed feature finite mixture model (MFMM) is formulated to be the estimation of mixture densities. Finally, the double geometric constraints for L-2-minimizing estimate (L2E) based energy optimization is formulated in order to calculate a reasonable position in a reproducing kernel Hilbert space. Extensive experiments on UAV images with different viewpoints are conducted. Experimental results show that our method provides better performances in most cases after comparing with six state-of-the-art methods.