基于聚焦合成的显微三维成像系统  

Three-Dimensional Microscopy Image System Based on Depth from Focus

作　　者：姜志国[1] 史文华[1] 韩冬兵[1] 孙维忠[1] 刘莉[1] 

机构地区：[1]北京航空航天大学,北京100083

出　　处：《CT理论与应用研究（中英文）》2004年第4期9-15,共7页Computerized Tomography Theory and Applications

基　　金：国家自然科学基金(编号:60372017);北航图像技术研发中心资金

摘　　要：本文利用显微镜物镜焦深范围有限,显微三维样本成像时需要一序列图像才能聚焦清晰特点,研究开发利用图像处理技术扩展显微成像焦深范围,应用改进Laplacian聚焦算子实现序列图像的融合显示,同时,提出了一种基于区域小波变换的序列显微图像融合算法并加以实现。用高斯插值算法得到三维显微图像的高度图,设计了一种网格轮廓线算法对高度索引图进行了三维重建显示和测量,并利用立体视觉原理生成了三维物体的左右视图和立体视图。应用此方法可以完成显微目标的表面三维参数的测量分析。Microscope suffers from the problem of limited depth of focus, which precludes the observation of complete image of a three-dimensional object in a single view. In this paper, extending depth of focus by multi-focus image fusion is investigated. The principle of image fusion based on depth of focus is analyzed, and one spatial domain multi-focus algorithm is implemented. Then a multi-focus micro-image fusion algorithm based on area wavelet transform is presented, which offer improved performance over existing algorithms when processing transparent images. In order to obtain 3D structures of the object, Gaussian interpolation algorithm has been selected to interpolate a small number of focus measure values to obtain accurate depth estimates. Based on the above DEM and Montage images, the 3D surface of micro-object is reconstructed by surface rendering algorithm. And the stereo pair images(left and right) and anaglyph image are also emulated.

关 键 词：聚焦合成 显微三维图像 聚焦深度 立体图像 

分 类 号：TH742[机械工程—光学工程]