Tracking Of Rat's Spontaneous Activity

The experiment of rat's spontaneous activity is a widely used method in pharmacological evaluation and zoological research.Nowdays,with the application of the new technologies,especially the video- track and image recognition technology,and invention of new devices in this field,the pharmacology and zoology scinence have been improved greatly.This dissertation mainly research detection and tracking technology of the mouses in the chamber.First,it describes current development status and signification of rat's spontaneous activity research,and then use an improved two-dimensional Otsu threshold segmentation algorithm in mouse image by researching the threshold segmentation technology. And we present a method of mouse detection based on weighted hierarchical histograms of oriented gradients in region of interest. Finally we use the algorithm based on parametrically deformable contour models to track the moving mouses. The main research innovations and contributions are summarized as follows:1. We take a deep research into the threshold segmentation algorithm , propose an improved two-dimensional Otsu threshold segmentation algorithm . This method is better to separate the mouses out from the background and lay a good foundation for further detection and tracking of mouses.2. Based on the traditional HOG & SVM detection algorithm,we present a method of mouse detection based on weighted hierarchical histograms of oriented gradients in region of interest.We achieve the HOG feature of the mouses in region of interest by weighted hierarchical histograms of oriented gradients and then use the SVM classifier to detect and identify the mouses.3. We explore the tracking method based on model and raise an algorithm based on parametrically deformable contour models according to the features of the rat's spontaneous activity in the chamber.we present a tracking system by this method and adding a supervision module in the system.Due to the supervision module, the system is able to recover from errors and successfully track objects for long time periods.The system also shows a good performance in many situations such as multi-target tracking, target under cross-blocked ,and so on.