| 摘 要: 细胞追踪算法的研究对建立细胞的生长发育模型并探索它的基因结构和功能至关重要。针对蜂窝状植物细胞结构复杂,在成像中存在严重噪声干扰,且采集的图像未经配准等难点,提出了一种深度局部图匹配网络(DLGM-Net)模型,利用局部图的拓扑结构信息和深度相似度信息寻找相似度最高的细胞作为种子细胞对,然后从种子细胞周围进行扩散匹配。为了进一步提高追踪精度,将DLGM-Net与多种子细胞对投票机制结合,纠正种子细胞对扩散匹配过程中出现的错误匹配。实验结果显示,所提算法能够有效地实现未配准图像下的细胞追踪,与现有局部图匹配算法相比,追踪精度平均提高了7.25%。 |
| 关键词: 植物细胞 深度局部图匹配网络 多种子细胞对 细胞追踪 |
|
中图分类号: TP391.41
文献标识码: A
|
|
| Multi-seed Cell Pair Tracking Algorithm Based on Deep Local Graph Matching Netwo |
|
LI Jieqin1 , XIE Dingfeng1 , ZHANG Te1 , WANG Xueping2
|
(1.College of Information Engineering, Hunan Industry Polytechnic, Changsha 410000, China;
2.College of Information Science and Engineering, Hunan Normal University, Changsha 410081, China)
lijieqin4568@163.com; coolboyxie@163.com; 376591767@qq.com; 1139590669@qq.com
|
| Abstract: Cell tracking algorithm is critical for modeling of the cell growth patterns and exploring the structure and function of their genes. In view of the difficulties of complex cell structure of honeycomb plants, serious noise interference in imaging, and unregistered images, a Deep Local Graph Matching Network (DLGM-Net) model was proposed, which used the topological information and depth similarity information of local maps to find the cells with the highest similarity as seed cell pairs, and then diffusion matching is performed from the seed cells. To further improve the tracking accuracy, the DLGM-Net is combined with a mult-i seed based majority voting scheme to rectify possible matching errors in the cell correspondence growing process. Experimental results show that the proposed algorithm can track cells effectively, and it gains 7. 25% improvement in tracking accuracy compared with the existing local graph model on average. |
| Keywords: plant cells deep local graph matching network mult-i seed cell pair cell tracking |
