一个基于知识的图象理解系统

基于知识的视觉系统有一定的现实意义。文章提出了一个基于知识的图象理解系统，在该系统中，结合景物知识进行区域信息与边缘信息的融合以得到分割结果，利用模型指导聚焦以减少特征提取的计算量，并使用多种模型匹配得到最终的场景解释。实验证明，文中构造的系统在系统的适应性、精确性和快速性方面有较大的改进，并取得了较好的解释效果。

A Knowledge Based Image Understanding System

Knowledge based vision system is very important in practice. In this paper,a knowledge -based image understanding system is presented. In the system,the segment result is obtained by fusing area information and edge information utilizing scene knowledge,the computation cost of feature extraction is reduced by using model to indicate focusing,and the final result is got through multi-model matching. The experiment results show that the system properties of precision,adaptability and rapidity are impoved,as well as the system can get better interpretation.