The latest research shows that 80 percents of the Japanese adult are infected by a periodontal disease. The pathological study on live periodontal bacteria has been not yet developed at a sufficient level for finding treatments and preventions against the periodontal disease. It is useful to design an automatic system that analyzes the live bacterial activity and dynamics in the mouth. This study presents an approach for analyzing the bacterial activity on a video sequence acquired by a differential phase microscope. By using a moving object segmentation algorithm, it has been successful to remove the background out of the video sequence and to detect moving bacteria. Bacteria are extracted by labeling connected pixels, and their locations and shapes are identified for tracking moving bacteria. The periodontal bacteria tracking system has been implemented on a digital signal processor (DSP) for real-time processing. The system offers two tracking modes: tracking all individual bacteria and tracking a single specific bacterium.