一种基于旋转不变性的角点检测方法研究  

Corner detection method based on rotation invariance

作　　者：吴敏杨 季锋 蒋明 

机构地区：[1]苏州科技大学数理学院,江苏苏州215009

出　　处：《液晶与显示》2018年第2期150-155,共6页Chinese Journal of Liquid Crystals and Displays

基　　金：国家自然科学基金(No.11172193)~~

摘　　要：针对显微立体视觉标定过程中,因显微成像产生的畸变和模糊而造成棋盘格标定板角点误检的问题,提出一种基于旋转不变性的角点检测方法,研究了棋盘格角点间的几何特征,对所有特征角点进行排序,通过实验建立数学模型,计算筛选的特征角点的置信度,准确提取出棋盘格角点位置,并通过一系列实验对这种方法的准确度、噪点数量以及精度进行验证。结果表明,基于旋转不变性的棋盘格角点检测方法,准确率达到91.2%,重投影误差的均方差约为0.1像素,具有较高的准确性和精度。Aiming at the error detection of the chessboard corners caused by distortion and blur of microscopic imaging during the calibration processing of the micro stereovision,a corner detection method of chessboard based on rotation invariance is proposed,which studies the geometry of chessboard corners and sorts all the characteristic corners.Mathematical model is established through experiments,the confidence level of filtered characteristic corners are calculated,which extracts positions of the chessboard corners and validate the accuracy,the number of noise and the precision through a series of experiments.The results show that the accuracy rate of the corner detection based on rotation invariance reaches 91.2%,and the mean square error of the re-projection error is approximately 0.1 pixels,which has a comparatively high accuracy and high precision.

关 键 词：旋转不变性 角点检测 几何特征 置信度 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]