【摘要】 研究了一种基于显微图像的几何重心法亚像素提取算法,用于实现超低转速的在线测量。首先根据显微图像法将超低转速的转动测量转换为平动测量;然后利用提出的几何重心法的亚像素提取算法,解决了目前常用的曲线拟合法和梯度法的亚像素提取时间长、抗图像干扰能力弱的问题。几何重心法的核心是在整像素提取的基础上,在3×3的像素范围内,分别计算参考像素灰度值与提取的整像素灰度之差的指数值,在获取的3点指数值上,利用几何重心法提取出亚像素值。通过光栅基准平台微位移抓拍图像的验证,提出的几何重心法比曲面拟合法、梯度法用时短,平均用时<0. 2 s,提取的亚像素值的标准差<1/1 000像素,满足了超低转速在线测量的基本要求。更多还原

【Abstract】 To achieve the ultra-low rotating speed online measurement,a geometric center of gravity subpixel extraction algorithm based on microscopic image was proposed. Firstly,the measurement of ultra-low rotation speed was transformed into translational measurement according to the microscopic image method,the surface fitting method and gradient method which are commonly used at presentnot only take longer to extract sub-pixel but also have weak anti interference ability,thesub-pixel extraction algorithm of the geometric center of gravity method solved these problems. The core of the geometric center of gravity method was to calculate the index value of the difference between the reference pixel gray value and the integer pixel gray value within the range of 3 × 3 pixel on the basis of the integer pixel extraction,then extracted the sub-pixel value from the acquired 3-point index value by using the geometric center of gravity method. Through the verification of micro-displacement captured images by grating reference platform,the average time of geometric center of gravity method was less than 0. 2 s,which was shorter than the surface fitting method and the gradient method. The standard deviation of the extract subpixel value was less than one thousandth of a pixel,which conformed to the basic requirements of ultra-low rotating speed online measurement.更多还原