Person identification using automatic height and stride estimation

We present a parametric method to automatically identify people in monocular low-resolution video by estimating the height and stride parameters of their gait. Stride parameters (stride length and cadence) are functions of body height, weight, and gender Previous work has demonstrated an effective use of these biometrics for identification and verification of people. In this paper, we show that performance is significantly improved by using height as an additional discriminant feature. Height is estimated by segmenting the person from the background and fitting their apparent height to a time-dependent model. With a database of 45 people and 4 samples of each, we show that a person is correctly identified with 49% probability when using both height and stride parameters, compared with 21% when using stride parameters only. Height estimates for this configuration are accurate to within /spl sigma/=3.5 cm. This method works with low-resolution images of people, and is robust to changes in lighting, clothing, and tracking errors.