文档介绍：Improved Mean Shift Segmentation Scheme For
Medical Ultrasound Images
Peirui Bai 1 , 2, Caihong Fu 2, Maoyong Cao 2, Yan Han 1
1. National Key Laboratory of Electronic Measurement 2. College of Information and Electrical Engineering
Technology Shandong University of Science and technology
North University of China Qingdao, China
Taiyuan, China bprbjd@
Abstract—Medical ultrasound image segmentation is a critical most of puter vision task is in the multivariate
step for image analysis and measurement. In this paper, an case. For example, natural images and ultrasound images
improved scheme is proposed to e limitations of the always contain rich texture, and over-segmentation often
traditional mean shift algorithm in medical ultrasound image exists using the traditional mean shift approach. To e
segmentation application. Two aspects are taken into account to the disadvantage of the traditional mean shift, various
improve the traditional algorithm: adaptive gray scale improved scheme has been proposed. For example, data-
bandwidth selecting and threshold-based region merging. driven local bandwidth selection approach for multimodal and
Experimental results of simulated and clinical ultrasound images multivariate data was presented in [6], and changeable color
indicate that better classification of objects and background bandwidth and direct density