MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202611091013 A) filed by Manipal University Jaipur on July 27, 2026, for A System And Method For Forensic Surveillance Image Enhancement Using A Two-Stage Deep Learning-Based Denoising And Super- Resolution Framework.
Inventor includes Dr Arjun Singh.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: The present invention relates to a system for forensic surveillance image enhancement using a two-stage deep learning-based denoising and super-resolution framework, namely Forensic-Vision SRCNN, to improve the quality of low-resolution surveillance images and video frames. The system comprises a Deep Convolutional Neural Network (DnCNN) denoising module with a Super-Resolution Convolutional Neural Network (SRCNN) reconstruction module to generate high- quality forensic imagery while preserving evidential authenticity. The DnCNN stage removes sensor noise and compression artifacts commonly found in CCTV footage, while the SRCNN stage enhances spatial resolution and restores critical visual details. The system utilizes a hybrid loss function combining Mean Squared Error (MSE) and Structural Similarity Index Measure (SSIM) to optimize reconstruction accuracy and perceptual quality. Trained using BSDS500 and fine-tuned on the SCFace surveillance dataset, the proposed framework achieves superior image quality compared to conventional interpolation and standard SRCNN methods. The invention is particularly useful in forensic investigations, facial identification, license plate recognition, and surveillance-based criminal intelligence applications.
Disclaimer: Curated by HT Syndication.
