Shadow Elimination Algorithm for Foreground Based on Statistical Model
Tian Zheng · Journal of Chinese Computer Systems · 2013
The existence of the shadow is one of the key factors which impact the result of the object detection,detecting and eliminating the shadow become the important research area in the target detection.Aiming to the shadow problem during the object detection,we used the histogram to statistic and analysis the color feature of the shadow under the HSV color space,got the shadow feature of the H,S,V channels.Then established the Gaussian shadow model on each channel according to the statistic information.Based on the model had built,we proposed a novel algorithm to eliminate the shadow,using the foreground pixels to matche the model then determine and eliminate the shadow.Compared with the similar algorithms,the results show that the proposed algorithm can eliminate the shadow correctly in real-time under different scenarios,and it performs better on the metrics of shadow detection rate and the shadow discrimination rate.