摘要
针对光学零件表面划痕检测精度较低的问题,提出一种多尺度线检测与离散正交多项式曲线拟合相结合的亚像素边缘检测算法。首先,为克服噪声等干扰,基于不同尺度线检测特性,提出改进的多尺度线检测,并通过该方法得到划痕初始轮廓。然后,对边缘点及其邻域在梯度方向上的点进行筛选。最后,利用正交多项式曲线拟合具有模型简单、求解速度较快、拟合精确的特点,对所选取离散点进行曲线拟合,以实现亚像素边缘检测。实验结果表明,该算法检测精度较高,具有较强的鲁棒性,同时更好地适用于屋脊边缘检测;较最小二乘法时间大大降低,且低于较快速的三次样条插值法,平均计算时间比为19.9∶4.73∶1,处理效率显著提高,能够满足工程应用中快速、高精度检测要求。
Aiming at the problem of low precision in scratch detection on optical lens surface,we proposed a sub-pixel edge detection algorithm,which combines the multi-scale line detection and discrete orthogonal polynomial curve fitting. First,to overcome interference of noise,based on the characteristics of line detection in different scales we proposed an improved multi-scale line detection,and obtained initial contour of scratches by this method. Then,we screened the edge points and the points of their neighbourhood on gradient direction. Finally,by making use of the advantages of orthogonal polynomial curve fitting in simple model,fast processing and precise fitting,we fitted the selected discrete points into curve so as to realise the sub-pixel edge detection. Experimental results show that the proposed method has higher accuracy and robustness,and is suitable for roof edge detection better. In terms of processing time,the method is greatly reduced when contrasting with the least square method,and is also faster than cubic spline interpolation method,their average computation time ratios are1∶ 19. 9∶ 4. 73,the processing efficiency is significantly improved. This method can meet the requirements of fast and high precision detection in engineering applications.
出处
《计算机应用与软件》
CSCD
2016年第11期157-160,179,共5页
Computer Applications and Software
基金
四川省科技支撑计划基金项目(2014GZ0005)
关键词
光学器件
划痕检测
线性检测器
曲线拟合
亚像素边缘检测
Optical devices
Scratch detection
Linear detector
Curve fitting
Sub-pixel edge detection
