Statistical Approach to a Color-based Face Detection Algorithm EE 368 Digital Image Processing Group 15 Carmen Ng, Thomas Pun May 30, 2002
Statistical Approach to a Color-based Face Detection Algorithm Assumptions 4 Stages: Pre-processing Skin Color Region Labeling Statistical Face Selection Techniques Edge Detection Advantages/Disadvantages 11/21/2018
Statistical Approach to a Color-based Face Detection Algorithm Assumptions: Color image Multiple faces with similar area Face orientation 11/21/2018
Statistical Approach to a Color-based Face Detection Algorithm I . Image Pre-processing Boundary extension Improves accuracy 11/21/2018
Statistical Approach to a Color-based Face Detection Algorithm II . Skin Color Region Labeling Color-based Chrominance extraction in YCbCr space Morphological operations Dilation and erosion 11/21/2018
<= Original Image Rough Mask => 11/21/2018
Binary Mask after Morphological Operations 11/21/2018
Statistical Approach to a Color-based Face Detection Algorithm III . Statistical Analysis Popular area finder Facial feature detector (holes in binary images) Popular area, width and height Face rejection Reject unpopular areas 11/21/2018
Selected Face Regions after Stage III 11/21/2018
Statistical Approach to a Color-based Face Detection Algorithm IV . Facial Feature (Eye) Detection Approximate eye location LPF to remove noise Edge detection to locate strong edges 11/21/2018
After LPF and Edge Detection Typical Background Typical Face After LPF and Edge Detection 11/21/2018
Statistical Approach to a Color-based Face Detection Algorithm Results/Conclusions: 88% success rate Adv: fast, no training required,work with video compression std. DisAdv: min of faces required in image, work best with reliable facial detector 11/21/2018