摘要
针对经典不变特征提取算法时间较长的缺点,提出一种局部特征结合形状轮廓匹配的建筑物识别算法.首先根据整体建筑物容易受到旋转、倾斜度的影响,提取已知建筑物的局部特征点,确定其方向,对其位置、角度等进行矢量描述;同时依据建筑物不同状态下的尺度变化信息,制定任意形状轮廓匹配相似度准则及映射函数,实现不同光照、不同尺度下的建筑物识别.仿真实验结果表明,该方法可以快速、准确地适应不同环境下的建筑物识别.
In order to overcome the shortcomings of the traditional feature extraction algorithm,a building recognition algorithm based on local feature and shape contour matching is proposed.First of all,according to the influence of the rotation and inclination of the whole building,the local feature points of the buildings are extracted,the direction determined,and its position and angle described by vectors.At the same time,according to the scale change information in different states of the building,the similarity criterion and mapping function of arbitrary shape contour matching are formulated to realize the recognition of buildings with different illumination and different scales.The simulation results show that the proposed method can help identify buildings quickly and accurately in different environments.
出处
《西安建筑科技大学学报（自然科学版）》
CSCD
北大核心
2017年第5期752-756,共5页
Journal of Xi'an University of Architecture & Technology(Natural Science Edition)
基金
陕西省自然科学基础研究计划项目(2016JM6079)
陕西省教育厅专项科研项目(14JK1429)
关键词
城市建筑物识别
局部特征
轮廓匹配
尺度信息
urban building recognition
local feature
contour matching
scale information
