Development of the application of mathematical morphology operations in remote sensing image processing

Aim of project
The aim of the project is to improve the theoretical basis and applications of mathematical morphology in digital image processing (especially satellite images processing). The four main scientifi c aims of the project can be distinguished.

The first aim is to develop algorithms for distinguishing heterogeneous classes of land cover in the remote sensing images. Popular classification methods are based on the similarity of digital numbers of pixels and do not take into account contextual information, so they are very efficient in classification of relatively homogeneous classes (like water, bare soils or meadows) but inefficient in classification of heterogeneous classes (like orchards or built-up areas). Morphological operations due to its unique possibility of modifying the image depending on its structure can be used in distinguishing heterogeneous classes of land cover.

The next aim is automatic generalization of raster images. Some mathematical morphology operations show an important potential in this matter. Applying a proper operator with the structuring element properly chosen, an efficient generalization, i.e. suppression of unimportant information and preserving important one, will be possible.

The next aim is to test the efficiency of image filtration using morphological operators. The unique characteristics of mathematical morphology show important potential in this matter.

An additional aim of the project is to create software (free license) allowing the application of the developed algorithms.

Expected results
Development of algorithms improving the efficiency of image (especially satellite and aerial) classification. Development the algorithms for automatic generalization of raster images. Creating of free software allowing the application of the developed algorithms.