Tomonaga Sachihiko, Baba Misuzu, Yamazaki Sadao, Baba Norio
Major of Informatics, Graduate School, Kogakuin University, 2665-1 Nakano, Hachioji, Tokyo 192-0015, Japan.
Research Institute for Science and Technology, Kogakuin University, 2665-1 Nakano, Hachioji, Tokyo 192-0015, Japan.
Microscopy (Oxf). 2014 Oct;63(5):357-69. doi: 10.1093/jmicro/dfu021. Epub 2014 Jun 17.
We devised a new field-of-view autotracking method for online tomography reconstruction based on a cross-correlation between a pair of neighbours, called back-projected ray images, among a specimen tilt sequence. One ray image is calculated through normal filtered back-projection only in the cross-sectional plane from each projection image. This ray-image matching can reliably track the field-of-view because a pair of neighbouring ray images mostly cross-correlates at the existing three-dimensional object position. Online experiments using real specimens resulted in successful autotracking performance with high accuracy, and online tomograms were obtained immediately after the final tracking at the last tilting angle.
我们设计了一种新的视野自动跟踪方法,用于基于在线断层扫描重建,该方法基于标本倾斜序列中一对相邻图像(称为反投影射线图像)之间的互相关。仅在每个投影图像的横截面平面中通过常规滤波反投影计算一个射线图像。这种射线图像匹配可以可靠地跟踪视野,因为一对相邻的射线图像大多在现有的三维物体位置处互相关。使用真实标本的在线实验获得了高精度的成功自动跟踪性能,并且在最后倾斜角度的最终跟踪之后立即获得了在线断层图像。