No abstract

Color features can improve the performance of object tracking under various lighting conditions, such as light sources, shadows and highlights. The same object under different lighting conditions can have dissimilar appearances in images recorded by the same camera. Therefore, color invariant features which are robust to lighting condition changes are desired for object recognition and tracking. In this paper, the color space selection for shadow detection in casually captured scenes is addressed. The performance of shadow detection can be improved significantly through appropriate color space selection strategy. Hence, several well known color spaces are experimented. It demonstrate that the selection of color space is important for shadow detection and extraction processes. The experimental results on real-life scenes show that proposed color space is efficient over other color spaces.

This paper presents a novel method for skin segmentation in color images using piece-wise linear bound skin detection. Various color schemes are investigated and evaluated to find the effect of color space transformation over the skin detection performance. The comprehensive knowledge about the various color spaces helps in skin color modeling evaluation. The absence of the luminance component increases performance, which also supports in finding the appropriate color space for skin detection. The single color component produces the better performance than combined color component and reduces computational complexity.

No abstract

No abstract