基于贪心策略的视觉SLAM 闭环检测算法

Abstract

动态环境与视觉混淆严重影响视觉闭环检测性能.基于贪心策略,提出了一种在线构建视觉词典的闭环检测算法.算法优先处理Surf描述与已有单词Surf描; 述欧式距离最大的特征点,改进特征点与单词Surf描述最近邻的约束条件,生成了表征性能强、量化误差小的视觉词典,算法具备实时性,并在动态环境图像集; 与视觉混淆多发生的图像集上,在确保100%,准确率的条件下,最大召回率分别提升了5%,与4%,.

The performance of loop closure detection is seriously affected by; dynamic objects and perceptual aliasing in the environment.Based on; greedy strategy,a real-time loop closure detection approach using online; visual dictionary is proposed.The process of dictionary construction; gives priority to dealing with Surf feature that has the maximum; Euclidean distance from the closest vocabulary word.A more; discriminative and representative visual vocabulary is produced through; adding constraint condition to the nearest neighbor distance.This visual; vocabulary guarantees a small quantization error.The proposed approach; meets real-time constraints.Experiments based on datasets from dynamic; environments and visually repetitive environments demonstrated that the; largest recall rate increased by 5%, and 4%, respectively at 100%,; precision.