Restoration of Corrupted Region and Segmentation

Abstract

Image segmentation is fundamental to many image analysis problems. It aims to partition a digital image into a set of non-overlapping homogeneous regions. This paper describes a new method for restoration and segmentation in corrupted text images on the basis of color feature analysis by tensor voting in 3D. It is show how feature analysis can benefit from analyzing features using second order tensor. Proposed technique is applied to text images corrupted by manifold types of various noises. Firstly, selected dominant features in color space are analyzed by tensor voting in 3D, and noises are removed by an adaptive vector median iteratively. Finally, the region segmentation is performed by adaptive mean shift and separated clustering method respectively. We present experimental results of the proposed method operating on an image corrupted by various noises.