A Novel Method and Apparatus for Perceptually Optimized Natural Image Repair
Computing & Wireless : Other Computing
Available for licensing
- Alan Bovik, Ph.D. , Electrical and Computer Engineering
- Anush Moorthy , Electrical and Computer Engineering
Current digital cameras need manual tuning in order to set at least some of the parameters that control image capture. Other parameters do not seek to maximize the perceptual quality and/or the naturalness of captured image.
The invention allows for automated tuning of these parameters to maximize perceptual quality/naturalness.
The invention is a novel algorithm that can automatically control camera parameters using perceptual quality features that are uniquely derived from natural scene statistics; e.g., models of natural (non-synthetic) images. Thus, the invention can simultaneously ameliorate any one or more of a set of camera setting-related distortions on a given image.
- Unprecedented perceptual quality of imaging
- Can be measured in real time
- Continuous assessment of quality
- Continuous identification of distortions and their severity
- Distortion reduction in a perceptually optimal manner
Photography and communications
Proof of concept
- 1 U.S. patent issued: 9,277,148