论文部分内容阅读
亚像素定位技术是实现基于视觉的高精度MEMS测量的关键。文章提出了一种混合非线性优化和离散傅里叶变换的亚像素定位方法。首先通过相位相关法估计出像素级平移参数,然后利用互功率谱的整体信息,在其邻域范围内进行上采样矩阵傅里叶变换,精确定位相位互相关谱的峰值,测得亚像素级平移参数,最后实现了图像的高精度匹配。实验表明,该算法的定位精度优于扩展的相位相关法,像素定位精度能达到0.01像素,能够满足MEMS运动图像的高精度定位,且具有更好的噪声抑制能力和计算复杂度较低的特点。
Sub-pixel positioning technology is the key to enabling vision-based, high-precision MEMS measurements. In this paper, a hybrid sub-pixel localization method based on nonlinear optimization and discrete Fourier transform is proposed. Firstly, the pixel-level translation parameters are estimated by the phase-correlation method. Then, using the overall information of the cross-power spectrum, the up-sampling matrix Fourier transform is performed in the neighborhood of the cross-correlation spectrum to accurately locate the peak of the phase cross-correlation spectrum. Translate the parameters, and finally achieve high-precision image matching. Experiments show that the proposed algorithm is superior to the extended phase correlation method in positioning accuracy. The pixel positioning accuracy can reach 0.01 pixel, which can meet the high-precision positioning of MEMS moving images with better noise suppression and lower computational complexity .