Automatic aerial image shadow detection through the hybrid analysis of RGB and HIS color space
暂无分享,去创建一个
This paper presents our research on automatic shadow detection from high-resolution aerial image through the hybrid analysis of RGB and HIS color space. To this end, the spectral characteristics of shadow are firstly discussed and three kinds of spectral components including the difference between normalized blue and normalized red component - BR, intensity and saturation components are selected as criterions to obtain initial segmentation of shadow region (called primary segmentation). After that, within the normalized RGB color space and HIS color space, the shadow region is extracted again (called auxiliary segmentation) using the OTSU operation, respectively. Finally, the primary segmentation and auxiliary segmentation are combined through a logical AND-connection operation to obtain reliable shadow region. In this step, small shadow areas are removed from combined shadow region and morphological algorithms are apply to fill small holes as well. The experimental results show that the proposed approach can effectively detect the shadow region from high-resolution aerial image and in high degree of automaton.
[1] Peng Gong,et al. Integrated shadow removal based on photogrammetry and image analysis , 2005 .
[2] Arthur R. Weeks,et al. Practical considerations on color image enhancement using homomorphic filtering , 1997, J. Electronic Imaging.
[3] Touradj Ebrahimi,et al. Cast shadow segmentation using invariant color features , 2004, Comput. Vis. Image Underst..
[4] Bui Tuong Phong. Illumination for computer generated pictures , 1975, Commun. ACM.