A Novel Beard Removal Method based on Structural Similarity and Co-ordinate Transformations

摘要

Face reconstruction from available images of a human face is a vital step to achieve various objectives, such as study of human anatomy, intrusion detection, face recognition, content based retrieval, and law enforcement. This paper introduces a novel approach for removal of beard from bearded images using the concept of structural similarity and Co-ordinate transformations. The proposed method utilizes the inherent characteristics of structural similarity in human faces which may be present in the form of similar eye-sockets, chin, lips, nose etc. The reconstruction is based on the observation that two humans having high similarity in some facial feature (such as eye sockets) are likely to have similarity in other features (say chin structure) as well. The technique works efficaciously on different type of faces, as it searches similar faces from database and achieves the desired goal of beard removal by accomplishing complicated task of chin reconstruction on the basis of structural mapping in two or more images.