| 生物荧光谱分离端元提取算法的实现与比较 |
| |
| 引用本文： | Chong M,Qin B. 生物荧光谱分离端元提取算法的实现与比较[J]. 中国医疗器械杂志, 2010, 34(4): 258-262. DOI: 10.3969/j.isnn.1671-7104.2010.04.006 |
| |
| 作者姓名： | Chong M Qin B |
| |
| 作者单位： | 上海交通大学生命科学技术学院生物医学工程系,上海,200240 |
| |
| 基金项目： | 国家自然基金项目,国家973项目,上海交通大学PRP项目 |
| |
| 摘 要： | 选取和实现了PPI（Pure Pixel Index）和N-FINDR两种端元提取算法,在荧光染料发光成像的实验条件下对算法进行了对比发现：N-FINDR计算速度快,能准确计算出最优m个端元的优点,但需要对m数目预先准确设定,有可能会忽略本是端元的像素点;PPI算法计算速度较慢,会找出一些并不是端元的素点,但同时不会因为m的误差而误删除端元;本研究进一步发现,先通过一个较少循环的PPI算法来剔除大量的像素点,然后再通过N-FINDR算法对于剩余的像素点进行计算处理,可以更快速准确地分离出我们所需要的端元点。
|
| 关 键 词： | 生物荧光成像 端元分离 PPI N-FINDR |
Comparison of endmember extraction at spectral unmixing for biological fluorescence imaging |
| |
| Chong Minqi,Qin Binjie. Comparison of endmember extraction at spectral unmixing for biological fluorescence imaging[J]. Chinese journal of medical instrumentation, 2010, 34(4): 258-262. DOI: 10.3969/j.isnn.1671-7104.2010.04.006 |
| |
| Authors: | Chong Minqi Qin Binjie |
| |
| Affiliation: | Biomedical Engineering Department, School of Life Science &Techndogy, Shanghai JiaoTong University, Shanghai 200240. |
| |
| Abstract: | We applied two classical algorithms in hyperspectral imaging to extract endmember for biological fluorescence imaging. A combined algorithm was found to initialize the PPI with decreasing the number of multi-spectral pixels, with subsequent N-FINDR refinement, which could make a better result. |
| |
| Keywords: | PPI N-FINDR |
| 本文献已被 维普 万方数据 PubMed 等数据库收录！ |
|
