Pheromone prediction-based ant colony multi-cell tracking system

Abstract

The invention discloses a pheromone prediction-based ant colony multi-cell tracking system. The system comprises four modules: a pheromone field prediction module, an ant decision module, a pheromonefield Gaussian fitting module and a state estimation module. Through analysis of the modules, multi-cell tracking is realized. After an original picture is input, based on a cell pheromone field Gaussian fitting result of a previous frame, pheromone field prediction is performed by utilizing a Gaussian mixture model; a pheromone field is constructed by utilizing a bell curve-based pheromone diffusion model in a pheromone gradient-based ant working mode; and through a K-mean clustering method, cell position estimation is performed after removal of a false target caused by clutters, and finallyassociation is performed by utilizing cell distance features to obtain cell motion tracks, thereby realizing the multi-cell tracking.

Description

technical field [0001] The invention belongs to the field of cell tracking, and more specifically relates to an ant colony multi-cell tracking system based on pheromone prediction. Background technique [0002] Cells are the basic unit of life activities. For the embryonic development, evolution and life maintenance of any organic life, the proliferation, differentiation and migration of cells are essential links. Therefore, research on the analysis of cell behavior is very valuable in many fields, including stem cell research, tissue engineering, drug development, genetics and proteomics, etc. Medical image processing is an important branch and research direction of computer vision research. The use of computer for image processing reduces human interference, can reduce the workload of medical workers, and improve the accuracy of analysis. Therefore, extracting target feature information and target trajectory from cell image sequences is an important and basic task in med...

AI Technical Summary

Problems solved by technology

When a cell adheres or even overlaps with another cell, the target may not be lost in the process, but when the target is separated due to similar cell characteristics, the tracking may switch, and cell label switching occurs.

Therefore, adhesion or even overlap will greatly increase the switching rate of cell labels, which will bring great difficulties to the reliability of target tracking.

[0005] 2. When the speed and direction of cell migration change suddenly, due to the loss of part of the target information, it is mistakenly thought that the cell has disappeared, resulting in unstable tracking or even loss of cells

[0006] In recent years, the automatic tracking methods proposed by scholars rely on the quality of cell image detection and segmentation, and the accuracy of cell tracking in complex situations is not high, and it is difficult to accurately describe its statistical characteristics.

However, the mature multi-target tracking technology for complex situations such as dense targets and occlusions in video images is affected by the particularity of the shape difference between cells in the microscopic image sequence and the similarity between cells and background gray values. Cannot be directly extended to the field of cell tracking

At present, there is still a lack of effective research theories and methods for tracking adherent cells in cell image sequences.

Images