Intelligent Skin Color Model Selection for Face Detection
0 likes428 views
This document proposes an algorithm for detecting faces in images using skin color modeling and analysis. It generates nine skin color models from face image data in different color spaces. A statistical threshold is applied to the skin color models to detect skin pixels. Spatial and morphological filtering enhances the targeted face images. K-means clustering is then used to localize multiple faces by cropping targeted regions represented by 4-neighbor ellipses. The algorithm was implemented successfully using local face databases and a projection-based method to select the best performing skin color model for face detection.
Intelligent Skin Color Model Selection for Face Detection
- 1. 1 Intelligent Skin Color ModelIntelligent Skin Color Model Selection for Face DetectionSelection for Face Detection Setiawan Hadi, Adang Suwandi A, Iping Supriana S, Farid Wazdi Universitas Padjadjaran, Bandung, Indonesia Institut Teknologi Bandung, Indonesia IntroductionIntroduction • Face detection is a preprocessing step of facial recognition system (Essential) IntroductionIntroduction • Goal: localize face(s) in digital image and/or in real time video Our Research ApproachOur Research Approach • Skin-based face detection • Skin color is represented in three color space (rg, HSB and YCbCr) • Using nine skin color models, generated mathematically from various face images • Apply statistical-based detection threshold for skin detection • Use projection-based approach for evaluation criteria of skin model selection • Implement spatial and morphological filtering approach for enhancing face image • Using k-means for multiple face localization in image and apply 4-neigbourhood ellipse representation for cropping the targeted face • Using local face databases for experiment
- 2. 2 General FrameworkGeneral Framework Face DatabasesFace Databases For Generating Skin ModelsFor Generating Skin Models Skin Color ModelsSkin Color Models Histogram of generatedHistogram of generated skin color modelskin color model
- 3. 3 Statistics of Skin ModelsStatistics of Skin Models ThresholdingThresholding FilteringFiltering Filter SettingsFilter Settings
- 4. 4 SkinSkin DetectionDetection Algorithm w h e re P s k in ( i, j ) is p ro b a b ility o f p ix e l P a s s k in p ix e l if in c lu d e d in d is t r ib u t io n s k in m o d e l D M k fo r e v e ry c o lo u r s p a c e s R n . Pskin(i, j) = Pskin(i, j) ∈ DMk ∀ P(i, j) ∧ ∀ Rn Numerical result of detectionNumerical result of detection ProjectionProjection--based Detectionbased Detection
- 5. 5 KK--MeansMeans ResultResult ConclusionConclusion • A combined algorithm for detecting faces in an image has been proposed and successfully implemented using multiple face image • A simple evaluation criteria for skin model selection, based on image profiling in horizontal and vertical projection, has been experimented. • Nine skin models have been explored and used in the experiment for selecting the best model that give the best face detection result. • Several preprocessing steps in image processing such statistical thresholding using empirical and Chebyshev’s rules, filtering using sand and pepper noise filtering have been implemented and can be used for enhancement of the targeted image • Face localization technique based on intelligence k-means clustering algorithm has been implemented successfully. Intelligent Skin Color ModelIntelligent Skin Color Model Selection for Face DetectionSelection for Face Detection Setiawan Hadi, Adang Suwandi A, Iping Supriana S, Farid Wazdi Universitas Padjadjaran, Bandung, Indonesia Institut Teknologi Bandung, Indonesia