This project has the aim to describe and to develop techniques to segment images with fine structures, and also to analyse and classify these images. Because of the fragility of this kind of structure, computer vision principles should be used besides of image processing techniques. In addition, new segmentation quality measures for this kind of application should be developed. Some of the potential applications of these techniques are the segmentation of images with plant roots and neurons, images of retina and also the segmentation of rivers and roads in satellite images. Mathematical models should be used to model the fine and brunched structures of the images making it possible to develop these techniques. Our motivation is to create a new set of techniques sensible to this kind of structure, which is not much explored in the literature. The project embraces the joint use of image processing and analysis techniques and computer vision. Furthermore, it has a wide range of real applications.
News published in Agência FAPESP Newsletter about the scholarship: