An Enhanced Computer Vision By Using MLP Approach To Forensic Face Sketch Recognition System‎

Abstract

Technologies for suspect identification, detection, and recognition have become more critical in recent years. As a result, face recognition is an almost commonly used biometric technique. Investigators for Criminal and forensic computer vision researchers are interested in the human-recognized face sketches were drawn by artists. Hand-drawn face sketches are, according to studies, ‎still extremely rare, both in terms of artists and number of drawings, since forensic artists ‎prepare victim drawings based on descriptions were provided by eyewitnesses following an incident‎. Masks are sometimes used to conceal standard facial features such as noses, eyes, lips, and skin color, but face biometrics' outliner features are impossible to conceal. This paper concentrated on a particular face-geometrical feature that could calculate some similarity ratios between composite template photos and forensic sketches. Computer vision techniques such as Two-Dimensional Discrete Cosine Transform (2D-DCT) and the Self-Organizing Map (SOM) Neural Network are used to design a system for composite and forensic face sketch recognition.