利用物体结构信息进行影像线特征关系匹配
Image Line Feature Relationship Matching with Object Structural Information
-
摘要: 针对影像线特征匹配的复杂性,给出了根据物体结构信息,将影像匹配问题转化为确定图同构问题的匹配策略。首先,利用影像特征提取获得的点、线特征建立描述物体结构信息的关系表,通过对左右不同影像的图形进行区域分割获得候选面特征,利用矢量数据拓扑关系的自动建立算法得出左右影像关联图的弧段-多边形拓扑文件。然后,根据基匹配及拓扑关系表,利用图深度遍历的方法进行匹配传递,按面、线、点的顺序完成不同影像间对应物体同名线的匹配。由于关系表的一致性表明了物体构成特征要素的一致性,可使整个影像的匹配搜索范围变小,从而提高影像匹配的准确性与匹配速度。Abstract: Aiming at the complexity of image line feature matching,we propose a matching strategy turning to the establishment of object topology relation table of describing object structure based on the object structure information.The method takes advantage of the graphical relations of image point and line features obtained by image feature extracting to build object topology relation table of describing object structure.Candidacy polygon features are acquired through segmentation geometry graphics of left and right image.The arc-polygon file of association graph of left and right image is obtained using vector data topology relation automatism algorithm.Then,the transmission match is carried out by the method of depth-first traversal based on base match and topology relation table.As the relationship tables consistency shows the composition of objects,features elements of consistency,the matching search scope of whole image is made smaller,thereby the image matching accuracy and speed is enhanced.