Robust subspace clustering by combined use of knnd metric and svd algorithm.pdf

Robust Subspace Clustering by Combined Use of kNND Metric and SVD Algorithm Qifa Ke and Takeo Kanade Computer Science Department, Carnegie Mellon University {ke+,tk}@cs.cmu.edu Abstract Subspace clustering has many applications in com- puter vision, such as image/video segmentation and pat- tern classification. The major issue in subspace clustering is to obtain the most appropriate subspace from the given noisy data. Typical methods (e.g., SVD, PCA, and Eigen- decomposition) use least squares techniques, and are sen- sitive to outliers. In this paper, we present the k-th Nearest Neighbor Distance (kNND) metric, which, without actually clustering the data, can exploit the intrinsic data cluster structure to detect and remove influential outliers as well as small data clusters. The remaining data provide a good initial inlier data set that resides in a linear subspace whose rank (dimension) is upper-bounded. Such linear subspace constraint can then be exploited by simple algorithms, such as iterative SVD algorithm, to (1) detect the remaining out- liers that violate the correlation structure enforced by the low rank subspace, and (2) reliably compute the subspace. As an example, we apply our method to extracting layers from image sequences containing dynamically moving ob- jects. 1 Introduction Subspace clustering has many applications in computer vision, such as image and video segmentation [18], layer segmentation [12], and pattern classification (see [7]). In these approaches, the input high-dimensional data are pro- jected or mapped onto a low dimensional subspace where the clusters become tighter and easier to identify. The major issue of subspace clustering is to obtain the most appropriate subspace from the given noisy data. Tra- ditional approaches to subspace estimation, such as Sin- gular Value Decomposition (SVD), Principal Component Analysis, and Eigen-Decomposition, are based on the least squares technique, and are therefore sensitive to outliers in the