Image Segmentation by semantic method

作者:

Highlights:

摘要

The problem of region detection is addressed. Linear and quadratic approximation schemes are used to approximate the regions in an image. A set of attributes, which represent the properties of a region, are defined. A distance function, which has structural as well as semantic part, is introduced. This distance function is used to decide whether a group of pixels in an image forms a region. A number of image models formed from the combination of various approximation schemes and attribute sets are studied. These image models are tested on a number of real world examples, and a promising set of attributes and approximation schemes for region detection are reported. The results of region detection on a number of images using the proposed image models are presented.

论文关键词:Region detection,Distance function,Structural distance,Semantic distance Attributes

论文评审过程:Received 29 May 1986, Revised 19 December 1986, Available online 19 May 2003.

论文官网地址:https://doi.org/10.1016/0031-3203(87)90077-X