Abstract:
In a previous work, we proposed a Comparative Morphology (CM) construction scheme that generalized Pytyev's Morphological Image Analysis approach onto a wider range of practical image comparison applications. Within a Guided Contrasting framework, a filtering procedure and a related change detection algorithm were developed. In this work, we propose a class of CM filtering in which Mathematical Morphology operators introduced by Serra are used as smoothing operators that offer monotonically non-increasing (non-decreasing) filtering, in contrast to linear diffusion filtering and non-linear median filtering. The results of experiments on change detection based on the new CM filtering are discussed in comparison with other morphological procedures.