期刊名称：International Journal of Image Processing (IJIP)
电子版ISSN：1985-2304
出版年度：2013
卷号：7
期号：3
页码：302-313
出版社：Computer Science Journals
摘要：This paper proposed the active contour based texture image segmentation scheme using the linear structure tensor and tensor oriented steerable Quadrature filter. Linear Structure tensor (LST) is a popular method for the unsupervised texture image segmentation where LST contains only horizontal and vertical orientation information but lake in other orientation information and also in the image intensity information on which active contour is dependent. Therefore in this paper, LST is modified by adding intensity information from tensor oriented structure tensor to enhance the orientation information. In the proposed model, these phases oriented features are utilized as an external force in the region based active contour model (ACM) to segment the texture images having intensity inhomogeneity and noisy images. To validate the results of the proposed model, quantitative analysis is also shown in terms of accuracy using a Berkeley image database.
关键词：Linear Structure Tensor; Quadrature filter; Active contour; Image Segmentation
