Abstract
In this paper, we consider the problem of subspace clustering for image data under occlusion and gross spatially contiguous noise. The state of the art subspace clustering methods assume that the noise either follows independent Laplacian or Gaussian distributions. However, the realistic noise is much more complicated and exhibits different structures in different scales. To address this issue, we propose a multi-scale framework that extracts a clean self-expressive dictionary through an iterative approach and is capable of identifying probable corrupted elements in each sample. Using this information, not only we can estimate parameters of each subspace more accurately but also by optimizing a matrix completion problem based on group sparsity, we can recover corrupted regions more precisely and hence achieve higher clustering accuracy for corrupted samples. Numerical experiments on synthetic and real world data sets demonstrate the efficiency of our proposed framework in presence of occlusion and spatially contiguous noise.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.