Adaptive Multiresolution Filter Based Multiview Face Recognition

Free registration required

Executive Summary

Face images under wild atmosphere suffer from the changes of heterogeneous factors such as camera view, illumination, expression, etc. Tensor analysis brings a trail of analyzing the influence of heterogeneous factors on facial variety. However, the Tensor Face model develops an obstacle in symbolizing the non-linearity of view subspace. In this paper, to snap this limitation, the authors furnish a adaptive multi-resolution based technique which contains a view manifold-based Tensor Face in which the latent view manifold preserves the local distances in the multi-view face space. Moreover, a Kernelized Tensor Face for multi-view face recognition is named to freeze the structure of the latent manifold in the image space.

  • Format: PDF
  • Size: 4839.79 KB