SAFETY IN PUBLIC AREAS USING FACE RECOGNITION

Abstract

Face recognition systems are generally used for security purposes. Face recognition is a biometric method that compares a person’s face that is live capture image or a video source with the stored images. Protecting the people from the criminals in public areas and at the same time accounting for law enforcement is a great challenge. Facial recognition is used for the very purpose. For any facial recognition system, there is a primary step that is ought to be done. Face detection is the preliminary step for face recognition. There are many algorithms and classifiers for face detection. Some of them are Segmentation algorithm, PCA, SVM, AdaBoost etc. Face detection itself is not sufficient. The detected face need to be accurate in various aspects like when the faces are rotated, under lighting conditions, complex backgrounds, variety of skin tones etc. These became challenging factors in face detection process. Many systems were proposed earlier for face detection. These existing systems ensure face detection in color images or background complexities or lighting conditions. Some common issues in these systems is that they have to go through many stages repeatedly leading to more time consumption, noisy output, less accuracy and efficiency levels. Hence this paper proposes a system for detection and recognition using Haar – cascade. Haar-cascade classifies the input into frames which for feature extraction. The proposed system detects faces in a group of people, variability in skin tones, variability in scale, in the presence of outliers. The recognition system recognises a suspect by comparing the face of the suspect with the faces that were stored in the database. The suspects face is automatically forwarded to nearby police station to catch the suspect. The proposed system provides a fast detection rate accounting for better accuracy and efficiency levels. This paper provides safety to the public in crowdy areas like railway stations, bus stands, market areas, shopping malls, parks and so-on.