Subspace techniques for image understanding

A unifying framework for solving several problems in image understanding is proposed based on the exploitation of subspace fitting techniques originally developed for antenna array processing applications. The SLIDE (subspace-based line detection) algorithm is a framework for introducing the concept of subspace fitting into problems in image understanding and computer vision. The basic formulation of SLIDE provides a solution to the problem of multiple line fitting in binary or gray-scale images. However, as is shown in this paper, its modified versions can be applied to several other image analysis problems. >