面向在线测量的亚像素提取与实验验证

Research on improved sub-pixel extraction algorithm for real-time detection

为解决亚像素提取时间过长以及解决不同材质的灰度图像对亚像素提取精度影响的问题,提出了一种几何重心法的亚像素提取算法。几何重心法首先通过相关分析获取整像素值,在整像素值附近对3×3的相关值进行几何差值,再通过重心算法实现对亚像素的快速提取。为验证几何重心法,采用光栅微移动平台对不同摄像头距离、不同背景光、不同材质的灰度图条件进行了亚像素对比实验,实验结果表明,相比于传统曲面拟合法和梯度法,几何重心法在亚像素提取时间、抗干扰能力以及误差等方面都具有明显的优势,特别是亚像素提取时间仅为其他算法的1/5或更少,满足了在线测量的实时要求。

In order to solve the problem of long sub-pixel extraction time and the influence of gray images with different materials on the precision of sub-pixel extraction,a sub-pixel extraction algorithm based on the geometric center of gravity method is proposed.Firstly,the geometric center of gravity method obtains the whole pixel value by correlation analysis method,then obtains the geometric difference value by the correlation value of 3×3 near the integer pixel value,and fast sub-pixel extraction is realized by gravity center algorithm.In order to verify the geometric center of gravity method,the grating micro-mobile platform is used to carry out the sub-pixel contrast experiment on the grayscale images with different camera distances,different background lights and different materials.The experimental results show that compared with the traditional surface fitting method and gradient method,the geometric center of gravity method has obvious advantages in the aspects of sub-pixel extraction time,anti-interference and error,etc.In particular,the sub-pixel extraction time is only 1/5 or less than the other algorithms,which meets the requirements of real-time online measurement.