Detecting of different kinds of changes in multitemporal remote sensing images. Two Multitemporal images, acquired over the same area at different times, are independently classified and land-cover transitions are estimated according to a pixel-by-pixel comparison. Unsupervised methods available in preceding pixel-by-pixel comparison methods are not effective in handling the images. In many real applications, it fails to collect ground truth information for multitemporal images and also loss of change information should takes place. To overcome the unsupervised framework is designed for detecting the changes in remote sensing images. Two different data sets having different property representations allow to easily display and understand changed information. Here, the information about multiple kinds of changes can be preserved as well as it is easy to visualize those images. It makes the process a better one in detecting maximum number of changes from the multi temporal image. The plan is to exploit the potentialities of the proposed technique in the context of more complex approaches to change detection like those exploit multi scale/multiresolution information intrinsically present in VHR images and the ones robust to registration noise.