Conventional methods to detect changes between temporal images are subject to the effect of illumination variance and registration noise. The edge structure information in image was employed in this paper to detect changes. A new conception based on biological vision principle, called Edge Tag, was introduced to describe the edge structure. After the correlation calculation between two Edge Tags, decision rule was made to locate the outline of changed area. The Edge Tag based change detection was robust to illumination variance and registration noise. Experiments on simulated data and real images verified its superiority.