應用獨立成份分析法於真人偵測

Application of Independent Component Analysis in Liveness Detection

人們可以很容易的辨識真人的臉孔或是照片的臉孔，甚至是人臉的表情變化、嘴部的移動、頭部的旋轉、眼睛的變動。然而，這些辨識對電腦計算而言是相當複雜的，甚至會受限於環境條件。本研究透過一創新方法的描述與實作，評估一運用低階的網路攝影機所攝取之影像，在進行量測該影像之心率時面臨的防偽問題。藉由應用獨立成份分析法於影像的彩色頻道，萃取臉部區域血流脈動，獨立成份分析法會選出涵蓋的最大能量成份，接著再以此作進一步的分析。本研究發展一方法去區別真人及欺騙樣本的血流脈動峰值波形訊號，此方法也許能對於之後偵測人物之可適性增強有所助益。

Human is able to distinguish a live face and a photograph without any effort, since human can very easily recognize many physiological clues of liveness, for example, facial expression variation, mouth movement, head rotation, eye change. However the tasks of computing these clues are often complicated for computer, even impossible for some clues under the unconstrained environment. This study describes, implements and evaluates a novel methodology for anti-spoof from measurements of the cardiac pulse rate extracted from video recorded by a basic webcam. By applying independent component analysis (ICA) on the color channels in video recordings, we extracted the blood volume pulse (BVP) from the facial regions. The component whose power spectrum contained the highest peak was then selected for further analysis. This study develops a methodology to perform the BVP peaks in the waveform signal from liveness (positive samples) and spoofs (negative samples). This proposed methodology may have a great value in monitoring person after adequate enhancements are introduced.