3D chrominance histogram based face localisation. (25th March 2011)

• Record Type: Journal Article
• Authors: Beniak, M.; Pavloviþová, J.; Oravec, M.
• Is Part Of: International journal of signal and imaging systems engineering. Volume 4:Number 1(2011)
• Page Start: 3
• Page End: 12
• Publication Date: 2011-03-25
• Subjects: face detection -- face localisation -- YCbCr colour space -- 3D histogram -- FERET image database -- gradient projections; Signal processing -- Periodicals; Imaging systems -- Periodicals; Information theory -- Periodicals; 621.3822
• Languages: English
• ISSNs: 1748-0698

Abstract

The paper describes face detection and face features localisation. We use YCbCr colour model because of chrominance components separation. The model is compatible with the video-broadcasting applications and videophone technology. First, we create a colour model for skin colour segmentation, and in the next part, we apply the created model to skin colour detection. Our approach is composed of pixel classification based on chrominance components and mathematical morphology. The utilisation of 3D histogram for chrominance components, instead of separated chrominance components, upgrades algorithm efficiency. In the face regions, we apply intensity and gradient projections to localise the face features.