Fast scale invariant multi-view face detection from color images using skin color segmentation & trained cascaded face detectors

Face detection is important step in face recognition, expression analysis, security, surveillance which has challenges due to multiple scales, views, rotations of faces & false background objects. Skin color segmentation, connected component extraction & correlation analysis on image is done to reduce search space & to improve detection rate. Cascaded face detectors are trained using Viola & Jone's Adaboost based Machine learning algorithm for each possible range of views & possible rotations. Segmented regions of 16*16 sizes are given to Cascaded face detectors to verify the presence of face. Experimental results show that it has very good detection rate for frontal & remarkable rate non-frontal, multi-view faces with negligible time duration in poor background/weather/lighting conditions.