Shadow elimination approach based on region statistics and YUV color
LI Peng-hong · Journal of Optoelectronics·laser · 2009
A robust moving shadow elimination method is proposed based on region statistics and YUV color space.Firstly,the method uses the neighbor pixel illumination linearity between shadow and background to estimate the background region illumination after image partitioning.Secondly,the difference image is achieved between blend GUASS model background and the estimated illumination background.Then,a shadow detection rule is designed by weighted variance statistics of the difference image.Finally,by using difference of chroma in YUV Color space and illumination change invariability of neighborhood shadow pixel,the fault shadow pixels are corrected.The experimental results of different video show the performance of the algorithm.The method also can eliminate the foreground hole.