基于金字塔和模糊聚类的路面图像拼接方法

Method of Pavement Image Splicing Based on the Pyramid and Fuzzy Clustering

公路图像具有特征复杂,特征点分布不均匀,噪声较多等特点,这对图像拼接带来了很多困难.针对以上特点,本文提出了一种基于图像金字塔、Harris角点检测和模糊聚类相结合的公路图像拼接方法.通过对原始路面图像进行多尺度、多分辨率的金字塔变换,形成几组尺度空间的分层结构,以该分层结构为基础,对每层结构进行Harris角点检测,并将得到的角点进行聚类和匹配,实现了相邻公路图像的精确拼接.

Pavement images usually have large noise and anomalistic distribution of characteristic points, which brings great difficulty for splicing. In view of the characters, this paper forms one kind of the method of road images splicing which based on image Pyramid, the Harris vertex examination and the fuzzy clustering. We form several groups of the structure of the criterion spaces carrying on the transform of the multi-criteria and multi-resolution Pyramid. Then we take this stratification structure as the foundation, and carry on the Harris vertex examination to each structure. Thus we make the characters carry on the cluster and realize splicing of neighboring road images.