Multiscale region segmentation of images using nonlinear methods

Region segmentation is the process of identifying the regions within an image, where a region is a group of connected pixels with similar properties. In this paper, we present a method of region segmentation using the scale-tree obtained from a datasieve - a recursive non-linear morphological filter. An initial segmentation of the image is done using the scale-tree based region growing method. We show that the properties of the datasieve scale-tree, in conjunction with the scale-tree based region growing method can be used to obtain the features of the objects in the image. The initial segmentation is then followed by post-processing methods for yielding a final segmentation. The results presented for several color images show the methods to be promising.