MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641077342 A) filed by Jagadeesan Sofia Priyadharshini; and Rajeev Gandhi Memorial College Of Engineering And Technology Autonomous on June 23, 2026, for Optimized Stochastic Computing Architecture Using Clock-Gated Mean And Square Root Units.
Inventors include Dr. J. Sofia Priya Dharshini; D. Manasa; V. Yeswanth; and K. Praveen Kumar.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: Low-light image enhancement is one of the major issues in the computer vision domain; low light image enhancement is needed across the various application domains to include: surveillance, autonomous driving, and mobile photography. Currently, zero-shot- low-light enhancing techniques yield poor results due to low-light conditions along with the absence of guidance being provided by the image's contents. This new zero- shot-low-light image enhancement algorithm will not utilize structured detail or illumination. Most recent low-light image enhancement techniques use paired datasets or multiple sources of supervision (i.e., guided diffusion with textual and CLIP features). In contrast to these approaches, we will replace the multi-modal guidance with a physically-based interpreted method (to be defined) for illumination, which can then be used to regulate the processes used to recreate the illumination and details of the input image. The current paper describes a new zero-shot enhancement model for low-light images, which is based on the concept of decomposing an image into illumination and detail through the use of a Fourier wavelet transform. This zero shot enhancement model will enhance the brightness of the illumination component of an image, while preserving structural information and removing noise. The model provides support for both the global and local (just in time) restoration of the illumination and detail components of the image. The framework of this novel approach has been extensively validated in various contexts.
Disclaimer: Curated by HT Syndication.
