SLAM系统中的自适应双阈值Canny算法  

A Self-adaptive Dual-threshold Canny Algorithm in SLAM Systems

作　　者：梁景辉 谢诗伟 陈畅频[3] LIANG Jinghui;XIE Shiwei;CHEN Changpin(School of Computer Science and Cyber Engineering,Guangzhou University,Guangzhou,China,510006;School of Mechanical and Electrical Engineering,Guangzhou University,Guangzhou,China,510006;Experimental Center,Guangzhou University,Guangzhou,China,510006)

机构地区：[1]广州大学计算机科学与网络工程学院,广州510006 [2]广州大学机械与电气工程学院,广州510006 [3]广州大学实验中心,广州510006

出　　处：《福建电脑》2024年第6期8-14,共7页Journal of Fujian Computer

基　　金：基于线性密钥共享方案的最优安全多方计算协议及应用国家自然科学基金(No.12171114);广东大学生科技创新培育专项资金资助项目(No.pdjh2023b0410);广州大学省级大学生创新训练项目(No.s202211078108)资助。

摘　　要：传统Canny算法在使用高斯滤波进行采样时,获得的滤波结果通常呈现为模糊图像,并且需要手动设定双阈值参数,因此缺乏对不同环境下的自适应性。为解决这个问题,本文提出一种SLAM系统中的自适应双阈值Canny算法,即采用双边滤波增强Canny算法的去噪和边缘特征保留能力,并通过自适应阈值选择算法实现在不同环境下确定最优的边缘特征提取阈值。仿真对比实验的结果表明,该算法在去噪和保留图像特征两个方面均具有更强的能力。The traditional Canny algorithm,when using Gaussian filtering for sampling,usually produces blurred images and requires manual setting of dual threshold parameters,thus lacking adaptability to different environments.To address this issue,this paper proposes an adaptive dual threshold Canny algorithm in SLAM systems,which enhances the denoising and edge feature preservation capabilities of the Canny algorithm through bilateral filtering,and determines the optimal edge feature extraction threshold in different environments through an adaptive threshold selection algorithm.The results of simulation comparative experiments indicate that the algorithm has stronger capabilities in denoising and preserving image features.

关 键 词：视觉同步定位与地图构建 边缘特征 双边滤波 自适应阈值 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]