摘要： 以往的单片建筑物检测,常假设相片是下视或近乎下视成像的,从而可忽略建筑物的透视形变。该文将长度约束与高斯映射和向量叉积相结合,提出了长度优先的向量叉积消失点检测方法,试验表明,该方法对分辨率较高且透视效果明显的图像,有着较好的检测效果。

关键词: 建筑物检测, 消失点, 模糊控制, 特征组合

Abstract: Initial approaches of monocular building detection often assume that all images are acquired with nadir or near-nadir imaging geometry, that perspective effects can be ignored. This article combines length constrains with Gaussian mapping and vector cross product, presents to a length-preferential vector cross product vanishing point detection method. Experiments show that this approach is suitable for the building detection of high definition and oblique images.

中图分类号: 

• TU198