Application of Preprocessing Techniques in Facial Recognition

Abstract

Identifying and recognizing criminals at a crime scene can be a complex and time-consuming process. Criminals may be identified through various methods such as fingerprints, DNA analysis, CCTV footage, or eyewitness testimony. The use of images captured by security cameras, along with fingerprint and DNA matching, requires access to a pre-existing database for effective recognition. Similarly, systems designed for human identification, such as those used for access control or attendance tracking, also rely on image capture and a database for accurate identification. This article discuss the methods for recognizing noised human faces by analyzing their features. Since images are multidimensional and can be affected by external factors that impact their clarity, creating an effective recognition model is a challenging task. To improve the system's accuracy, preprocessing techniques and feature extraction methods are applied to convert images into pixel-based data. The processed data is then fed into a Convolutional Neural Network (CNN) for classification and recognition. The study also examines the impact of three different preprocessing techniques and a comparative study of their effectiveness.