一种表达图像地物边缘的线段序列检测方法
A line segment sequence detection method for expressing image object edge
刘娜 1戴激光2
作者信息
- 1. 鹤壁能源化工职业学院应急管理系,河南 鹤壁 458030
- 2. 辽宁工程技术大学 测绘与地理科学学院,辽宁 阜新 123000
- 折叠
摘要
针对地物边缘提取中准确度与完整度难以兼顾、线段间缺乏连接关系的问题,提出一种表达图像地物边缘的线段序列检测方法.分析线段序列的构建规律,通过边缘分组界定地物边缘范畴,利用相位约束、距离约束、共线约束、连接性约束,解决遮挡或其他地物干扰导致的线段断裂问题,采用相邻性约束和端点共享约束对同边缘组内不同线段进行分类,同时对符合端点共享条件的线段进行更新,以构建线段序列.利用所提方法分别对街景、航空、卫星图像进行地物边缘检测实验,评价结果显示线段序列精度可达 98%,表明所提方法能够准确表达地物边缘,提升图像地物边缘表达的能力.
Abstract
In response to the problem of difficulty in balancing accuracy and completeness in edge extraction of ground objects,and lacking the connection relationship between the line segments,this paper abandons the idea of simply improving the integrity of the line segment,yet from the perspective of the same ground object edge,and proposes a line segment sequence detection method that expresses the edge of the image feature.The construction law of line segment sequence is analyzed,edge grouping to determine the edge category of the same ground object is used to solve the line segment fracture caused by occlusion or other ground object interference through the analysis of distance constraint,phase constraint,collinear constraint and connectivity constraint,and adjacency constraint and endpoint sharing constraint is used to classify different line segments in the same edge group.At the same time,the segments that meet the endpoint sharing conditions are updated to complete the construction of line segment sequence.Using the proposed algorithm,edge detection experiments are conducted on street view,aviation,and satellite images.The evaluation results show that the accuracy of the line segment sequence can reach 98%.The experimental results show that the proposed method can accurately express the edges of objects and improve the ability to express the edges of objects in images.
关键词
线段序列/地物边缘/端点共享/线段分类Key words
line segment sequence/object edge/endpoint sharing constraint/segment classification引用本文复制引用
基金项目
国家自然科学基金项目(42071428)
国家自然科学基金项目(42071343)
出版年
2024
国家科技期刊平台
NSTL
万方数据