Blurry image? Not a problem!
Problem being addressed
Low-rank matrix recovery from partial observations has been intensively studied due to its vast applications in computer vision and machine learning. For example, human facial images can be approximated very well by a low-dimensional linear subspace, therefore, a corrupted facial image can be recovered under the hypothesis that all the images lie in a low-dimensional subspace.
An optimization problem using a combination of low-rank and total variation regularizers; hence, it is expected to capture both, spatially local and global features, of the image better than if only one of the regularizers is considered. Towards this, the researchers regularize a recovery problem in such a way that allows them to reconstruct local information of an image as well as global information.
Advantages of this solution
Extensive experiments showed that the proposed algorithm outperforms state-of-the-art methodologies in recovering images.
Solution originally applied in these industries
Possible New Application of the Work
Image recovery algorithms are widely used in the art world and can shed light on the authenticity of certain objects as well as better understand their timeline.
Recovery of images under partial or corrupted observations is important for the video surveillance system and may play a crucial role for the police and security services.
Source URL: #############