Differential Geometric Approach to Change Detection Using Remotely Sensed Images

Change Detection using multi-temporal satellite images of same area is an established as well as actively pursued research problem. Most of the change detection techniques use algebraic or transform methods to do a pixel by pixel comparison of change detection. These techniques heavily depend upon the correct choice of threshold value to segregate the real changed pixels from the apparent changed ones. Also all these techniques can only compute the two dimensional change of the terrain surface from remotely sensed data. In this paper we propose a differential geometry approach to detect changes from remotely sensed images, which can detect the change using the geometric property of the pixels with respect to its surroundings. It can compute and filter the changed pixels having high curvature from that of flat (2D) changed pixels.