Image Denoising with Unsupervised, Information- Theoretic, Adaptive Filtering (UITAF)

The technology offers unsupervised adaptive filtering technique that automatically discovers the statistical properties of the signal and thereby reduces noise in a wide spectrum of images and applications. It is a nonparametric approach and adapts to the statistics of the input image. The new method improves the predictability of pixel intensities from the intensities in the neighborhoods by decreasing the joint entropy. The filtering operation is non linear and operates without any prior knowledge of the geometrical shape of the signal or noise.

Benefits
Various image filtering techniques, both linear and nonlinear, lack the generality to be easily applied to diverse image collections. Our technology is an efficient and generalized image denoising technique that can be applied to a wide spectrum of images and hence finds applications in a variety of image denoising applications.

Stage of Development
A provisional patent has been filed with the U.S. Patent and Trademark Office. This technology has been demonstrated to work in proof-of-concept experiments. It is available for developmental research support/licensing under either exclusive or non-exclusive terms.

Additional Info
*Suyash P. Awate, Ross T. Whitaker (2205) Higher-Order Image Statistics for Unsupervised, Information-Theoretic, Adaptive, Image Filtering. CVPR (2): 44-51.

Inventor(s): Suyash Awate, Ross Whitaker

Type of Offer: Licensing



« More Optical Science Patents

---