基于知识的图象分割方法

AN APPROACH TO KNOWLEDGE-BASED IMAGE SEGMENTATION

  • 摘要: 本文提出了一种基于知识,集分割与解释于一体的图象分割方法.着重研究了知识的表示、利用及区域解释的推导,设计了一种简捷、灵活的关系矩阵知识描述方法,并根据Waltz滤波算法的基本思想实现了一种快速的区域解释滤波方法.应用上述方法对室内景物进行了分割实验,结果表明,所提出的方法实现了对图象的完整分割与解释.

     

    Abstract: A new knowledg-based approach to integrating the segmentation and interpretation phases of sceneanalysis is presented.The representation of knowledge and its application,as well as the deduction of region interpretations are specifically studied.A brief and flexible knowledge description method usingrelational matrix is proposed.Based on the generalization of Waltz's filtering algorithm,a fast filteringprocess of region interpretation is performed.Experimental results from applying our approach to indoorscenes show that it is quite satisfactory.

     

/

返回文章
返回