Face identification at a distance with observation modeling

Face identification at a distance is a hard task. Images obtained from a long distance are often degraded by blurring or noise. This paper proposes a face identification method that uses a nonlinear optimum composite filter and subsequent testing stages. The composite filter detects a face region and identifies the face simultaneously. The composite filter is improved by applying the observation model to the training images. Subsequent stages consist of skin-color, edge-mask filtering, and similarity tests. Experimental and simulation results show that the proposed algorithm is effective in face identification.