摘要
提出了一种用于物体外形非接触式提取的综合立体匹配方法，该方法综合了基于面积的立体匹配过程和基于特征的立体匹配过程，能充分利用它们各自的优点，既能产生致密视差图，又能获得不连续处的准确匹配，解决了物体外形非接触式提取中的关键性的问题．
An integrated stereo matching method in non-contact extraction for object configuration is presented Unlike most stereo matching approaches,this method integrates areabased and featurebased primitives, which takes the advantages of the uniqueness of each of the previous.The method provides both the precise sparse disparity map and accurate location of discontinuities, and the dense disparity map for continuous surface The key problems in noncontact extraction for object configuration can be solved Good results are obtained in the applications for synthetic image and real object
出处
《中山大学学报（自然科学版）》
CAS
CSCD
北大核心
1998年第4期55-60,共6页
Acta Scientiarum Naturalium Universitatis Sunyatseni
基金
广东省博士后基金
关键词
计算机视觉
物体外形
非接触式
立体匹配
computer vision, stereo matching, stereo image pair, 3D reconstruction, extraction configuration
