Method for Identifying Individuals Using Lip Motion Features

Abstract

Speech speed is irregular in each speaker; however, it is important to adjust the structure of the speech data and allow for regional differences (accents or dialects) when identifying a person by their lip motion features. We propose a method to identify individuals by analyzing three kinds of lip motion features: lip width, lip length, and lip angle. The analysis is made on the basis of these features' relative values obtained from the primary and object frame. The proposed method's procedure involves three steps: automatic extraction of the lip motion features from the primary frame, change of the standardized data structure, and individual identification by fuzzy reasoning. We analyzed the lip motion features of 17 subjects using the proposed method, and achieved a successful identification rate of 100%. The results of the cumulative match rate (CMR) suggest that the proposed approach can be used to identify individuals more accurately than can DP matching.