摘要
为了有效提取特征点,提升误匹配点自动探测效果,提出一种基于图论的激光图像误匹配点自动探测方法。利用尺度不变特征变换检测激光图像极值点,完成特征点提取,归一化处理特征点,计算提取的特征点间的距离,按照距离为提取特征点构建激光图像完全图;建立激光图像导出图,通过迭代处理导出图,自动探测误匹配点。实验证明:该方法可有效提取激光图像特征点;在不同旋转角度时,该方法自动探测误匹配点的粗差误判率较低、内点比例较高,最高粗差误判率仅有6.8%,最低内点比例84.8%,说明该方法具备较优的误匹配点自动探测效果。
In order to effectively extract feature points and improve the automatic detection effect of mismatched points,an automatic detection method of laser image mismatched points based on graph theory is proposed in this paper.Using scale invariant feature transformation to detect the extreme points of laser image,complete the feature point extraction,normalize the feature points,calculate the distance between the extracted feature points,and construct the complete image of laser image according to the distance as the extracted feature points.Establish the laser image export map,export the map through iterative processing,and automatically detect the wrong matching points.Experiments show that this method can effectively extract the feature points of laser image,at different rotation angles,the gross error misjudgment rate of this method is low and the proportion of interior points is high.The highest gross error misjudgment rate is only 6.8%and the lowest proportion of interior points is 84.8%,indicating that this method has better automatic detection effect of mismatching points.
作者
唐滔
谭凤
TANG Tao;TAN Feng(Software College,Chongqing Institute of Engineering,Chongqing 400056,China)
出处
《激光杂志》
CAS
北大核心
2023年第5期210-214,共5页
Laser Journal
基金
重庆市教委科技项目(No.KJQN202001905)。
关键词
图论算法
激光图像
误匹配点
自动探测
尺度不变
完全图
graph theory algorithm
laser image
false match point
automatic detection
scale invariant
full graph
