Tracking neutrophil cells by active contours with coherence and boundary improvement filter

The present paper develops a new approach capable of tracking moving neutrophils in a video sequence. Each phagocyte moves such that its boundary mass center from the previous frame maps into the cell in the present frame. The frames are noisy, the neutrophil changes shape, has interior with low coherence and his boundaries are not clear. To track the cell, an active contour model is combined with coherence and boundary improvement approach, which uses eigenvalues and applies arithmetic mean filter on low coherence regions. Starting from the neutrophil´s boundary mass center projection, from the previous frame, the active contour enlarges toward the boundary of the cell in the present filtered frame. The boundary extracted form the filtered frame is mapped into its original. The paper ends with experimental results and comparison between the advantages and bottlenecks of the present study and works existing in the field.