With profunderization of peaceful city project, video investigation has played an increasingly crucial role in combating criminals and maintaining social stability. Across camera retrieval targeting specific passenger subject, also termed as passenger re-recognition, has become a popular topic in computer vision as one of the critical technologies. Breakthrough progresses have been made in the field including robust feature structure, scale learning and depth learning. Its best performance on public datasets has surpassed human recognition. However, most research environments for work are enclosed, with significant disparity from the open one for real application of video investigation. How should passenger re-recognition research be better tailored to demands in real application by being developed from enclosed environment to open world makes a promising trend for the field in future.