消除静态特征误匹配的机器人定位方法、存储介质及设备

本发明公开了消除静态特征误匹配的机器人定位方法、存储介质及设备，机器人定位方法以图像帧为输入，识别分割得到静态平面；静态平面区域中提取点面特征并进行点面特征匹配，采用双重特征约束的误匹配剔除策略；面特征的匹配首先构建出局部坐标系下相邻2帧中平面的匹配关系，利用基于平面几何约束的面特征误匹配算法去除匹配候选中的错误匹配，得到正确的面特征匹配关系以及点特征所属平面信息；根据点特征所属平面信息剔除非对应平面内的点特征匹配，再根据点特征位置剔除对应平面内的点特征误匹配。本发明保证特征匹配的正确率，提高机器人定位精度，从而构建更准确的轨迹图，并在动态场景中依然具有较高鲁棒性。 The invention...

Full description

Saved in:
Bibliographic Details
Format Patent
LanguageChinese
Published 28.04.2023
Subjects
CALCULATING
COMPUTING
COUNTING
IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
PHYSICS
Online AccessGet full text

Cover

More Information
Summary:本发明公开了消除静态特征误匹配的机器人定位方法、存储介质及设备，机器人定位方法以图像帧为输入，识别分割得到静态平面；静态平面区域中提取点面特征并进行点面特征匹配，采用双重特征约束的误匹配剔除策略；面特征的匹配首先构建出局部坐标系下相邻2帧中平面的匹配关系，利用基于平面几何约束的面特征误匹配算法去除匹配候选中的错误匹配，得到正确的面特征匹配关系以及点特征所属平面信息；根据点特征所属平面信息剔除非对应平面内的点特征匹配，再根据点特征位置剔除对应平面内的点特征误匹配。本发明保证特征匹配的正确率，提高机器人定位精度，从而构建更准确的轨迹图，并在动态场景中依然具有较高鲁棒性。 The invention discloses a robot positioning method for eliminating static feature mismatching, a storage medium and equipment, and the method comprises the steps: taking an image frame as input, and carrying out the recognition and segmentation to obtain a static plane; point-surface features are extracted from the static plane area, point-surface feature matching is carried out, and a double feature constrained mismatching elimination strategy is adopted; the matching of the surface features comprises the following steps: firstly, constructing a matching relation of planes in two adjacent frames under a local coordinate system, and removing wrong matching in matching candidates by utilizing a plane feature mismatching algorith
Bibliography:Application Number: CN202210964992