|
Cell movement plays a key role in many biological processes , analysis of cell movement , usually track the target cells . When the manual method of identification and tracking of millions of cells become impossible task , you can track and analyze thousands of cells automatic tracking system has become a research hotspot cell tracking . Cell tracking method based on graph theory is a new kind of cell tracking methods , is a multicellular while tracking the proposed method can take full account of the cells in the multi - cell image neighborhood topological relations . At a high level , this method is better than the previous method , there is a significant potential to achieve better tracking of a large number of cells , the method is worthy of further research . The multi - target tracking cell image sequence is the difficulty in the study of cell motility , cell movement in the high-density cell image complexity , this paper proposes a cell segmentation and tracking system . In cell division section for the experiment with a cell image sequence features , respectively, using the type of segmentation method . The basic cell division , there are some adhesion of cells because of the divided image obtained , in order to be separated , using the the cell contour tracking code based on the Freeman code method . Based on the analysis of the coding from the chain code features the outline of the cell morphology to identify adhesion cell image in the concave corner , then concave corner packet pairing is complete separation of cell adhesion . In the tracking part , the motion characteristics of the cells , the cells were divided into inert cells and non-inert cell area overlap method and topological constraints method for tracking . For high-density cell tracking , cell density and cell division can easily lead to the bug tracking problem , this paper proposed the the area item introduction of track topology constraints , to improve the original method , effective solution to this problem , helps to improve the tracking efficiency , especially for relatively large cell density and split the frequent cell image effect . In addition, this paper, according to the number of cells is set distance threshold , the algorithm more adaptable . This article has been two different image sequences tested and the results show that the proposed method has higher tracking accuracy .
|