Aerial and satellite images are useful and important data in the field of remote sensing. However, the shadows of clouds and buildings generally raise difficulties over various processing on images. In this paper, a comparative study on shadow detection in various color models is introduced. Without 3D geometry of scenes, a detection approach based on shadow properties is tested in various color models. The experimental results show that detecting shadow in the color models, HSV, YIQ and YCbCr, has better performance in terms of detection accuracy.