A Novel Hog-Based Template Matching Method for SAR and Optical Image
Deyu Song, Lan Liu, Xiangyin Zhang, Kaiyu Qin, Libo Wang · IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium · 2022
Due to multiplicative speckle noise in Synthetic Aperture Radar(SAR) image and significant intensity difference between different data, it is difficult to match SAR and optical image accurately. In this paper, we propose a novel template matching method for SAR and optical image named multi-Dimensional Matching Histogram of Oriented Gradient (mDM-HOG). Firstly, in order to reduce the negative effect of speckle noise on gradient calculation, the ratio of exponentially weighted averages(ROEWA) operator is introduced to calculate the gradient magnitude and orientation in SAR image. Then, using the obtained gradient information, we extract the 3-D pixelwise HOG feature for both images. Finally, we separate the 3-D feature map to nine sub-maps and measure the similarity of the sub-maps to obtain the template matching result. The experimental result shows that in comparison with the existing methods, the proposed template matching method has higher accuracy when locating the position of SAR image in optical image.