Description of Salient Features Combined with Local Self-Similarity for SAR Image Registration
- PDF / 2,830,423 Bytes
- 8 Pages / 595.276 x 790.866 pts Page_size
- 31 Downloads / 203 Views
SHORT NOTE
Description of Salient Features Combined with Local Self-Similarity for SAR Image Registration Lijuan Yang 1,2 & Zheng Tian 1,3,4 & Wei Zhao 1 & Weidong Yan 1 & Jinhuan Wen 1
Received: 26 April 2015 / Accepted: 11 April 2016 # Indian Society of Remote Sensing 2016
Abstract Local feature descriptor plays an important role in image representation and is helpful to further image processing. This paper proposes a local feature descriptor based registration method for synthetic aperture radar (SAR) images. The proposed method starts with identifying evenly distributed features by applying the divided salient image disk (SID) extraction method. To describe the shape content of local neighborhood, local self-similarity (LSS) descriptor is built in the local normalized region with a suitable size for every detected feature. Finally, the correspondence is found by measuring the similarity between LSS descriptors. The registration experiments on SAR images demonstrate that the proposed method can be applied to SAR image registration. Keywords Local self similarity . SAR . Image registration . Salient image disk
Introduction Image registration is a crucial step for remote sensing image processing. Due to the differences in image * Lijuan Yang [email protected]
1
Department of Applied Mathematics, Northwestern Polytechnical University, Xi’an 710129, People’s Republic of China
2
School of Science, Northwestern Polytechnical University, Dongda Chang’an, Xi’an, Shaanxi 710129, China
3
Department of Computer Science and Technology, Northwest Polytechnical University, Xi’an 710129, People’s Republic of China
4
State Key Laboratory of Remote Sensing Science, Beijing 100101, People’s Republic of China
mechanism, images captured by different sensors may appear different intensities and texture patterns. Especially for synthetic aperture radar (SAR) images, the existence of noise and deformable objects also makes image registration more difficult. So image preprocessing is essential before image analysis. Meanwhile developing robust image registration method has received more and more attention. In general, image registration methods can be classified into two categories, namely, feature-based methods and area-based methods (Zitova and Flusser 2003; Goshtasby 2005; Le Moigne et al. 2011). Feature-based methods firstly extract certain type of distinct feature (e.g., corner points, lines) using some interest point detector and then find the corresponding feature pairs based on the similarity of these features. Commonly-used feature detectors include Harris corner detector (Harris and Stephens 1988), Smallest Univalue Segment Assimilating Nucleus (SUSAN) (Smith and Brady 1997), Canny edge detector (Canny 1986) and so on. As one of the routine methods, harris corner detection is based on the pixel gradient of the image, which uses a covariant matrix of local directional derivatives to identify the locations in image with feature orientation varying significantly. For this kind of methods, the precise location o
Data Loading...
