3d Object Recognition From Appearance: Pca Versus Ica Approaches
IMAGE ANALYSIS AND RECOGNITION, PT 1, PROCEEDINGS(2004)
摘要
Two feature extraction techniques (PCA/ICA) for recognition of 3D objects from appearance are compared with respect to different recognition approaches (universal/object subspaces). A class separation ratio is defined, and several recognition experiments are performed using the COIL-100 database. The results show that both techniques produce similar recognition rates when universal subspaces are used; but, when object subspaces are used, ICA representation greatly outperforms the earlier PCA technique due to its ability to separate classes.
更多查看译文
关键词
Principal Component Analysis, Object Recognition, Recognition Rate, Independent Component Analysis, Training Image
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络