Abstract: To address the problem of poor robustness of pupil tracking in infrared images for gaze tracking systems, this paper presents a particle filter algorithm for pupil tracking based on pseudo-color map. This paper also presents a concept of triple-channel pseudo-color map (TCPCM) by use of the phenomena of bright and dark pupil and leads it into pupil tracking. TCPCM makes full use of the information of each channel. The color of pupil zone is distinct from that of other regions of face obviously, which will improve the stability and accuracy of tracking. A two-phase-updating method based on similarity measurements of histogram and geometry is adopted to improve the credibility of the particle weights. Fast feature extraction is introduced to improve system real-time performance, which not only increases the computational efficiency, but also enhances the robustness of the system. Results from an extensive experiment show a significant improvement of the proposed algorithm over existing eye tracking techniques for gaze tracking systems.