Skip to Main Content
Vision-based human identification is a reliable biometric system for determining the authenticity of a person's specific physiological or behavioral characteristic. Gait of a person is a non-invasive biometric characteristic that can be observed at a distance. In this paper, a novel method for improving the recognition accuracy is proposed by combining the isoluminance stereovision technique with the 3D template matching method. A 3D model consisting of depth information is constructed from stereo images using isoluminance lines for stereo vision. A canonical form of this 3D image is constructed and depth information is determined. These values are compared with a database of predetermined depth information for recognition purposes. The recognition accuracy is compared with HMM analysis performed on the same set of images. The results are similar and noted a slight improved accuracy for this proposed method. The images computed are samples stored in the Carnegie Melon University database.
Date of Conference: 4-7 Jan. 2005