Structural discrepancy search for images

We present structural discrepancy search algorithms for digital images. Our motivation is the problem of automatic object detection which (dis)appeared on (from) the space pictures of the terrain taken in different time spots. Based on the morphological image analysis by Pyt'ev we formulate general image structure concept and the corresponding structural discrepancy model. Using this framework search algorithms are given for grayscale and color images. Furthermore we conduct a comparative study for these algorithms using our special numerical technique both for adjusting their parameters and choosing the best algorithm. Also the software will be presented for this task.