19 September 2016 Object-based illumination normalization for multi-temporal satellite images in urban area
Author Affiliations +
Abstract
Multi-temporal satellite images acquisition with different illumination conditions cause radiometric difference to have a huge effect on image quality during remote sensing image processing. In particular, image matching of satellite stereo images with great difference between acquisition dates is very difficult for the high-precision DSM generation in the field of satellite photogrammetry. Therefore, illumination normalization is one of the greatest application technology to eliminate radiometric difference for image matching and other image applications. In this paper, we proposed a novel method of object-based illumination normalization to improve image matching of different temporal satellite stereo images in urban area. Our proposed method include two main steps: 1) the object extraction 2) multi-level illumination normalization. Firstly, we proposed a object extraction method for the same objects extraction among the multi-temporal satellite images, which can keep the object structural attribute. Moreover, the multi-level illumination normalization is proposed by combining gradient domain method and singular value decomposition (SVD) according to characteristic information of relevant objects. Our proposed method has great improvement for the illumination of object area to be benefit for image matching in urban area with multiple objects. And the histogram similarity parameter and matching rate are used for illumination consistency quantitative evaluation. The experiments have been conducted on different satellite images with different acquisition dates in the same urban area to verify the effectiveness of our proposed method. The experimental results demonstrate a good performance by comparing other methods.
Conference Presentation
© (2016) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Nan Su, Ye Zhang, Shu Tian, Yiming Yan, "Object-based illumination normalization for multi-temporal satellite images in urban area", Proc. SPIE 9976, Imaging Spectrometry XXI, 99760Q (19 September 2016); doi: 10.1117/12.2237254; https://doi.org/10.1117/12.2237254
PROCEEDINGS
6 PAGES + PRESENTATION

SHARE
Back to Top