New face segmentation technique insusceptible to illumination
MetadataShow full item record
This paper presents a robust face segmentation method dealing with the illumination problem effectively on color images. In color images, skin color is an important cue for face segmentation. Two sets of skin color data are acquired from Asian Database, i.e. normal conditioning skin pixels and illuminated skin pixels. Subsequently, both data are analyzed in YCbCr color space. As a result, proposed skin color normalization is implemented on the illuminated skin pixels with the intension of getting the skin pixel boundary as the priori knowledge. In order to remove the effect of illumination on images, Comprehensive Color Image Normalization, composing of row and column normalizations is applied on the input images. Row normalization is to reduce the lighting geometry while column normalization is to remove the color illumination. These normalized images are then compared to the predefined skin pixel boundary. Pixels that fall within the color boundary are denoted as skin color. In addition, elliptical skin boundary generates better performance than the rectangular boundary in the experiment. ©2008 IEEE.
Showing items related by title, author, creator and subject.
Li, B.; Xue, M.; Mian, A.; Liu, Wan-Quan; Krishna, A. (2015)In this paper we propose a robust face recognition algorithm for low resolution RGB-D Kinect data. Many techniques are proposed for image preprocessing due to the noisy depth data. First, facial symmetry is exploited based ...
Li, Billy Y.L. (2013)One of the most important advantages of automatic human face recognition is its nonintrusiveness property. Face images can sometime be acquired without user's knowledge or explicit cooperation. However, face images acquired ...
Duggal, R.; Gupta, A.; Gupta, Ritu; Mallick, P. (2017)© Springer International Publishing AG 2017. Convolutional Neural Networks (CNNs) are typically trained in the RGB color space. However, in medical imaging, we believe that pixel stain quantities offer a fundamental view ...