A Novel Hypothesis for Cast Shadow Modelling and Detection for Objects Tracking and Recognition

N. Al-Najdawi, H.E. Bez, and E.A. Edirisinghe (UK)


Shadow detection, Object Tracking, Image Transformation


Traditional computer vision applications are often chal lenged by the need to distinguish between objects and their shadows. Though, many shadow detection and re moval algorithms have been proposed in literature. Most of the proposed methods that claim to be object and en vironment independent include some minor assumptions about the scene geometry or spectral distributions of the light sources. In this work, based on a physically-derived hypothesis for shadow identiļ¬cation, novel, simple and fast shadow detection algorithms are proposed and imple mented in the spatial (Pixel) and frequency (Fourier) do mains. It is shown that the algorithms effectively remove shadows under various lighting and environmental condi tions. The proposed algorithms are able to detect shadows in both umbra and penumbra neighbourhoods .

Important Links:

Go Back