机构地区: 华南理工大学机械与汽车工程学院
出 处: 《仪器仪表学报》 2011年第2期426-431,共6页
摘 要: 针对光照变化、噪声、运动不连续及大位移量等影响基于计算机微视觉的微运动测量精度的问题,提出一种基于同态滤波的鲁棒多尺度微运动测量算法。首先采用同态滤波增强方法对显微视觉图像亮度不均匀进行了校正,并增强对比度,然后利用双权重函数,自动调节不同残差数据点的权重,去除残差过大的数据点,并采用多尺度金字塔由粗到精逐层迭代,精确地估计运动矢量。实验结果显示,新算法鲁棒性好,能有效地减弱光照不均匀的干扰,同时减少噪声和运动不连续而引起的界外值的影响,微运动测量精度达到0.01个像素。 Aiming at the problems that illumination variation,noise,motion discontinuity and large displacement may badly influence motion estimation precision in micro-motion measurement based on computer microvision,a robust multi-scale micro-motion measurement algorithm based on homomorphic filtering is proposed.Firstly,a homomorphic filtering method for image enhancement is used to correct uneven brightness and enhance contrast.Then biweight function is used to automatically adjust the weights of the data with different residual errors and remove the data with excessive residual errors;and a multi-scale pyramid is employed to accurately estimate the motion vectors through iterations gradually from coarseness to fineness.Experimental result show that the new algorithm has good robustness,it can effectively weaken the interference of uneven illumination and reduce the influence of outliers caused by noise and motion discontinuity;the accuracy of micro-motion measurement reaches 0.01 pixels.