Corrupted Region Restoration Using Second Order Tensors and Segmentation

In this paper, we address the problem of restoration efficiently and effectively, by formulating as one of surface inference from a sparse and noisy point set in a 3D space. Given a set of noisy data correspondence in corrupted images, our method extracts good matches and rejects noises. The methodology is unconventional, since, unlike most other methods optimizing certain scalar, objective functions, our approach does not involve initialization or any iterative search in the parameter space. Therefore, it is free of the problem of local optima or poor convergence. Subject to the general restoration of natural images, we detect corrupted region and noises by a novel computation scheme, 3D tensor voting, which is an instance of the more general tensor voting. In essence, the input set of matches is first transformed into a sparse 3D point set. Dense, 3D tensor kernels are then used to vote for the most salient surface that captures all inliers inherent in the input. lastly, density estimation for center modes detection and separated clustering algorithm is performed later for segmentation of values according to color components in the restored image. The experimental results show that proposed approach is efficient and robust in terms of restoring and segmenting corrupted color images.