Automatic Face Recognition

The system usually consists of four modules as shown in the figure: \ ref {fig: face_recogniton}: face detection, facial alignment, feature extraction and matching. In this article we focus on the focusing part, feature extraction and matching phases. \ begin {figure} [t] \ centering \ includegraphics [width = 5.5 in] {images / face_recognition} \ caption {face recognition processing flow. } \ label {fig: face_recogniton} \ end {figure} The face recognition system has been greatly improved since the 1980s.

Automatic face recognition is a complex task involving identifying faces and positioning facial features to identify faces. All faces have the same structure, but there are many environmental and personal factors that affect the appearance of the face. The main problem arising in automatic face recognition is the large variability of the recorded images due to gestures, lighting conditions, facial appearance, frontal images and the like. For images taken of different individuals at different times, there may be slight variations in facial expression and interpersonal variation. One way to overcome this problem is to include intra-individual changes in the training set image.

Based on pioneering research in key academic institutions in the United States and Europe, automatic facial expression recognition program was developed, so that it is widely used to detect faces, encode facial expressions, and recognize emotional state instantaneously became. The use of an inexpensive web camera eliminates the need for dedicated advanced equipment so it is suitable for automatic representation suitable for capturing face video in various natural environments such as respondents' home, workplace, automobile, public transportation, etc. Coding becomes possible.

Face Recognition System with Principal Component Analysis (PCA) Algorithm The automatic face recognition system attempts to find the identity of a given face image based on its memory. The memory of the face recognizer is usually simulated by the training set. In this project, our training set contains features extracted from the known facial images of various people. Therefore, the task of the face recognizer is to find the most similar feature vector among the feature vectors of a given test image in the training set. Here we identify the identity of the person whose image (test image) is passed to the system. In this project, PCA is used as a feature extraction algorithm. In the training phase, you need to extract feature vectors for each image in the training set. Let A be a training image of person A having a pixel resolution of M × N (M rows and N columns). The length (or dimension) of vector ÁA is M × N.