Natural Image Point Prediction Method
Computing & Wireless : Application Software
Available for licensing
- Wilson Geisler III, Ph.D. , Psychology
- Jeffrey Perry , Psychology
Characterizing the structure of natural images is critical for understanding visual encoding and decoding in biological vision systems and for applications in image processing and computer vision. The most studied statistical regularities in natural images involve first- and second-order (pair-wise) statistics. Further, it is typical to impose some invariance constraints in order to reduce dimensionality or otherwise simplify measurement of the statistical regularities. However, restricting measurements to pair-wise statistics and making invariance assumptions may miss important statistical structure. In order to move beyond these restraints and conduct measurements along higher numbers of dimensions than would otherwise be possible, researchers at The University of Texas at Austin have proposed a novel method for measuring moments along single dimensions, conditional on the values of other dimensions.
The proposed method is a demonstrable improvement from existing solutions to the problem of estimating original image pixel values given a digitized array of image pixel values. It does this by first measuring, with a novel direct technique, the average local statistics in space and/or time of natural images captured with an arbitrary image-capture device of interest, such as a digital camera. These statistics are used to create look-up tables that provide optimal Bayesian estimates of point (pixel) values.
- Can be used for a variety of image processing applications
- Faster and more accurate than standard methods
- Can be easily tailored to specific classes of image-capture device and image processing tasks, for even better performance
- Allows for good real-time performance
- Involves only estimated look-up tables as opposed to other methods
- Provides statistically optimal estimates given the local image properties considered
- The tables can be approximated with smooth functions to reduce storage.
Image processing (enlargement, deblurring, noise/artifact reduction and compression)