METHOD AND SYSTEM FOR AUTOMATED FACE DETECTION AND RECOGNITION

摘要

A method and a system for recognizing a person are provided to operate an intelligent video watching system and a biometric system precisely and stably by using various kinds of 3D(Dimensional) scenes, database lists, and training samples. A remote information collector transmits 2D images(101) captured by a 2D camera(102) to a face detecting/tracking device(103). The face detecting/tracking device tracks a face found in a frame during face search, face feature coordinate measurement, and video sequence processes. A face image quality evaluator(106) calculates a value of parameters included in the received image and transfers a result to a recognition mode selector(109). The recognition mode selector selects the most effective recognition mode for the current face image depending on a quality evaluation result. A biometric feature former(111) generates a simple shape of the face image processed by the selected recognition mode. A person recognizer(114) determines similarity between the generated shape and a template of a template database by measuring and summing up the similarity of various kinds of processed face images. A DBMS(DataBase Management System)(119) manages the template database storing the templates of the person included in a watch list.