Translator Disclaimer
22 May 2014 A stereo remote sensing feature selection method based on artificial bee colony algorithm
Author Affiliations +
To improve the efficiency of stereo information for remote sensing classification, a stereo remote sensing feature selection method is proposed in this paper presents, which is based on artificial bee colony algorithm. Remote sensing stereo information could be described by digital surface model (DSM) and optical image, which contain information of the three-dimensional structure and optical characteristics, respectively. Firstly, three-dimensional structure characteristic could be analyzed by 3D-Zernike descriptors (3DZD). However, different parameters of 3DZD could descript different complexity of three-dimensional structure, and it needs to be better optimized selected for various objects on the ground. Secondly, features for representing optical characteristic also need to be optimized. If not properly handled, when a stereo feature vector composed of 3DZD and image features, that would be a lot of redundant information, and the redundant information may not improve the classification accuracy, even cause adverse effects. To reduce information redundancy while maintaining or improving the classification accuracy, an optimized frame for this stereo feature selection problem is created, and artificial bee colony algorithm is introduced for solving this optimization problem. Experimental results show that the proposed method can effectively improve the computational efficiency, improve the classification accuracy.
© (2014) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Yiming Yan, Pigang Liu, Ye Zhang, Nan Su, Shu Tian, Fengjiao Gao, and Yi Shen "A stereo remote sensing feature selection method based on artificial bee colony algorithm", Proc. SPIE 9124, Satellite Data Compression, Communications, and Processing X, 912411 (22 May 2014);


Model-based feature extraction
Proceedings of SPIE (September 24 1993)

Back to Top