We propose a real-time key point tracking and mapping approach for continuous region tracking. The proposed approach is realized in a parallel way, which can be substantially optimized for multi-core processor. One thread tracks the key points employing our ultra-low-weight key point detector and hybrid SP-Tree, and the other thread handles the 3D mapping task. Experiments are conducted to demonstrate that the proposed approach can keep high accuracy while substantially reducing the computational load. In conjunction with background segmentation, the proposed approach can address region tracking task using PTZ Thermal Infrared Imager.