A significant challenge in fingerprint verification systems is their vulnerability to degraded image quality, which can lead to inaccurate and incomplete feature extraction, ultimately compromising system performance. The importance of assessing the quality and validity of captured fingerprint images cannot be overstated, as poor-quality images can have far-reaching consequences. In particular, false non-matches can have severe implications in negative recognition applications, such as watch list and duplication detection, where malicious individuals may intentionally tamper with their fingerprints to avoid identification. To address this issue, we propose a novel fingerprint identification and comparison approach based on the Hough transform. In high-stakes applications, individuals with malicious intent may deliberately alter their fingerprints to evade identification. Current methods rely on matching datasets by identifying the Rigid Core Delta point to facilitate fingerprint recognition. This paper introduces a novel approach to detect and correct skin distortion in individual fingerprint images. Our method employs a patch-based strategy, where rectangular regions are defined to identify areas of distortion within the fingerprint. By varying the patch size, distortion detection is improved, and a Support Vector Machine (SVM) classifier is trained to perform classification tasks. Furthermore, we treat distortion rectification as a regression problem, where the input is a distorted fingerprint image, and the output is a distortion-free field, enabling accurate fingerprint recognition.

FFT, SVM, classification,