用于双视图对应学习的分组置换增强网络  

Group Shuffle Enhancement Network for Two-View Correspondence Learning

作　　者：康政 赖桃桃 KANG Zheng;LAI Taotao(College of Computer and Information Sciences,Fujian Agriculture and Forestry University,Fuzhou,China,350028;School of Computer and Big Data,Minjiang University,Fuzhou,China,350108)

机构地区：[1]福建农林大学计算机与信息学院,福州350028 [2]闽江学院计算机与大数据学院,福州350108

出　　处：《福建电脑》2025年第2期12-17,共6页Journal of Fujian Computer

基　　金：福州市科技创新创业人才培育计划项目(No.2022-R-001)资助。

摘　　要：基于卷积神经网络的方法在估计运动场方面取得了进展,但在复杂场景中难以捕捉多个运动模式。为了解决这一问题,本文提出了分组置换增强块。通过将通道划分为多个子集并引入通道置换技术,促进子集间的特征交互,从而更精确地识别不同运动模式下的内点。实验结果表明,该方法在相对位姿估计任务上表现出更高的性能。The method based on convolutional neural networks has made progress in estimating motion fields,but it is difficult to capture multiple motion patterns in complex scenes.To address this issue,this paper proposes a group permutation enhancement block.By dividing the channel into multiple subsets and introducing channel permutation techniques,feature interaction between subsets is promoted to more accurately identify inliers under different motion patterns.The experimental results show that this method exhibits higher performance in relative pose estimation tasks.

关 键 词：双视图对应学习 分组置换增强 相对位姿估计 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]