FACEDETECTIONAPPROACHINNEURALNETWORK BASEDMETHOD FORVIDEOSURVEILLANCE

Neuralnetworks areadaptive information processing systems that offer attractive solutions forvideo surveillance. Thisapplication aimsat identifying particular patterns. Also,MPEG-4standard profiling strategy infacial animation guarantees that thestandard canprovide adequate solutions forvideo surveillance. The maingoal ofthis presentation istoprovide face detection forvideosurveillance usingneural network based method. After providing thecorresponding architecture forface detection, theemphasize isonthedetector which istrained withmultilayer backpropagation neural networks. Threedifferent facerepresentations aretaken intoaccount, i.e. pixel representation, partial profile representation andeigenface representation. Basedon this, three independent sub-detectors aregenerated. The detection rates aremeasured. Thecircle atabout 9400 indicates theposition wheretheneural network achieves theoptimal performance.