MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202641110077 A) filed by Vellore Institute Of Technology on September 14, 2026, for Method For Enhancing Latent Fingerprint Images With Explainable Attribution.
Inventors include Sudha M; and Chavan Ishita Himmatrao.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: ABSTRACT METHOD FOR ENHANCING LATENT FINGERPRINT IMAGES WITH EXPLAINABLE ATTRIBUTION A method for enhancing a latent fingerprint image (102) for forensic analysis comprises receiving the latent fingerprint image (102), preprocessing (104) the latent fingerprint image (102) to produce an input conforming to an input format of a trained model, and applying a trained conditional generative adversarial network (106) comprising a generator (108) and a discriminator (110) to generate an enhanced fingerprint image (112) in which ridge structure is restored. The method further comprises segmenting (114) the latent fingerprint image (102) into superpixels, generating perturbed images (116) by masking superpixel combinations, determining changes in enhanced outputs, assigning contribution scores (120) to superpixels based on the changes, and outputting a visual overlay (122) indicating superpixels most influential on the enhanced fingerprint image (112). (Fig. 1)
Disclaimer: Curated by HT Syndication.
