Color Segmentation and Color Correction Using Lighting and White Balance Shifts

摘要

A method was developed to segment an image foreground and background based on color content. This work presents an alternative to the standard blue-screen technique (weather man method) by exploiting the color shifts of light sources and filtering each camera lens and correcting white balances for each camera. we compressed the colors of the scene background into a chromaticity subspace to make the foreground-background segmentation easier to perform. The segmenta-tion is singular decomposition (SVD) based.