Color cast is a major problem of target detection and recognition base on image feature analysis. A color cast detection and correction algorithm for video-image in nature scene is proposed in this paper. To overcome the inaccurate detection in existed color cast detection method based on RGB space, the analysis method based on the color distribution characteristics of the equivalent circle in Lab space is proposed. In order to achieve better color correction results, the gray balance and perfect reflection method is combined to reproduce the actual color of the objects, especially a modified method for images with severe color deviation. Experimental results show that the proposed methods can detect the degree of color cast correctly, achieve better correction results, and meet the real-time requirements.