A Markov Random Field Framework for Finding Shadows in a Single Colour Image

Many computer vision algorithms, such as segmentation, tracking, and stereo registration, are confounded by shadows in images. Hence finding shadows in colour images is an important research issue. As opposed to the majority of techniques, which either need a sequence of images or require geometric information on images, this paper proposes an illuminant discontinuity measure by which shadow edges can be locally identified. We model the problem of finding shadows by a Markov random field which uses the new measure. The Markov random field provides a computational framework whereby local and area constraints can be optimized such that shadows can be segmented in a single colour image. Results are presented for real images and show accurate shadow extraction.