Research On Buildings Shadow Processing In High-Spatial Resolution RS Images

With the development of remote sensing technology and the improvement of satellite spatial resolution, remote sensing images are used more and more widely. However, remote sensing images are influenced by many factors in the process of imaging. Especially the tall buildings should be blocking of light that produce shadow in the image, which means the loss of image information and reduces image quality.Shadow is caused by lack of light radiation, it affects image recognition, feature extraction and classification and leads to influence of the precision of data processing, it even can not work sometimes. So, it is very important how to effectively reduce or eliminate the shadow in remote sensing image preprocessing.In general, the processing of shadow consists of two parts:shadow detection and shadow elimination.The main work of this paper is to study the mechanism and method of shadow in remote sensing image, in order to detect shadows accurately and eliminate shadows effectively.In this research, it analyzed the existing shadow detection methods firstly, and then conducted a comparative analysis of three common shadow detection algorithms based on multiband, HIS, RGB respectively. Shadow detection accuracy of experimental area image can reach above80%after using a combination of the accumulated detecting conditions in experiment. Secondly, in this paper, it took a study and analysis on existing shadow removal method, and achieved the shadow removal algorithm based on IHS space compensation. Besides, a shadow removal method based on classification is proposed and it extracts the all kinds of feature information in non-shaded area to make up for grayscale texture information corresponding to shaded area by the classification of shaded area and non-shaded area. Compared to compensation method based on HIS color space, it proved the feasibility and effectiveness of the method proposed in this paper. Finally, it finished the processing of the shadow in remote-sensing image by the plugin programming development on the software platform of OpenRS.