A statistic based approach for segmentation

Segmentation is one of the most active topics in image processing and computer vision fields. Although knowledge in this area is developing very fast [1][2 noisy images, especially the ultrasonic ones, remain difficult to process [3][4]. The main idea of this paper is to sort out the relevant changes of the local stationarity of the image with a statistical approach instead of a gradient based one [5][6]. One application case is reported.