Multispectral Image Matching Method Based on Histogram of Maximum Gradient and Edge Orientation
Quan Wu, Shipeng Zhu
Abstract
Motivated by the problem of nonlinear intensity changes in multispectral image matching, this letter introduces a robust and efficient image-matching method. The proposed method consists of three steps. First, control-point candidates are identified that are widely distributed in the areas of effective main structure. Then, a novel feature descriptor called the histogram of maximum gradient and edge orientation (HGEO) is proposed for the purpose of multispectral image matching. Finally, a bilateral matching process is carried out to perform the matching process and remove mismatches. The proposed method is successfully applied for matching various multispectral remote sensing images, and experiments are performed with typical datasets that are widely applied in tests of multispectral image matching. According to some popular feature descriptors, the test results demonstrate that the proposed HGEO achieves better matching performance than do many currently used methods.