摘要： 针对大规模显微图像的拼接，提出了一种显微图像拼接的全局对准模型。利用2D格状图表示图像的邻接关系，并根据局部对准的可靠性对格状图进行修正，以消除空白区的影响。根据图像投影坐标的误差建立全局对准模型，可以有效地消除误差积累。把系统偏斜校正作为约束条件加入全局对准模型，并利用非线性优化方法得到具有全局一致性的图像对准参数，成功地解决了大规模显微图像的拼接问题。

关键词: 图像拼接, 图像对准, 显微图像, 格状图, 偏斜校正

Abstract: This paper proposes a 2D grid graph based image mosaicing method for large-scale microscopic images. It uses 2D grid graph to represent the neighborhood relations of microscopic images, and modify the gird graph to eliminate the effect of blank areas according to the reliability of local alignment. It constructs the global alignment model with all the overlapping image pairs. It adds system skew rectification to global alignment model as the constrained condition. This paper gets globally consistent estimation of alignment parameters with nonlinear optimize method, and succeeds in the automatic mosaic construction of large-scale microscopic images.

Key words: Image mosaicing, Image alignment, Microscopic image, Grid graph, Skew rectification

中图分类号: 

• TP391