Facial recognition is one of the most important application in image processing. Face expression is recognized by emotions in our face. Clustering the faces based on the emotions of the person is achieved by finding the nearest neighbors using FLANN. Approximate rank order clustering algorithm is used to find the distance between similar images with accuracy and reduced run time. Clustering of faces is done by extracting the features of the images and the emotions are identified by landmarks extracted using DLib. Features which are used for clustering the face images are extracted using the Convolution Neural Networks (CNN). Emotions of a person are expressed as happy, sad, anger and neutral. Different landmarks for each expression is found and classified according to the corresponding landmarks. Accuracy in clustering of face images and emotion classification is compared using various datasets like LFW, ColorFeret and Frontal Images.
Volume 11 | 04-Special Issue
Pages: 1009-1026