A new real-time method for description and generalized segmentation of color images
K. I. Kiy · Pattern Recognition and Image Analysis · 2010
In this paper, a new method of automatic generalized segmentation of color images for real-time processing is described. The method is based on attaching to any image a new topological object, a geometrized histogram, which joins statistical and geometric image descriptions. The generalized segmentation is based on new clustering methods in the spaces of the geometrized histogram, which reduce the dimension of the problem, using the structures on the classifying space induced by the image itself. A software implementation provides processing real-world scenes, input from a video camera, at a high rate.