Abstract: In order to discover the leakage and crack diseases in tunnel in time, take early measures to reduce losses and save costs, a set of automatic tunnel inspection system based on multi-sensor was designed. By analyzing the characteristics of leakage water and cracks in tunnels, aiming at the phenomenon of leakage water, a multi-neighborhood dynamic anomaly region recognition algorithm was designed to detect leakage water by using infrared thermal imaging equipment to obtain pseudo-color image of temperature difference in the environment. Aiming at the problem that the cracks were small and difficult to identify, a high-resolution image was obtained by large-area array industrial camera, and a phase-based image was proposed. The image mosaic method of machine calibration obtained all the field of view in the tunnel, so as to expand the crack detection range. Then, the sensor module was installed in the tunnel motion inspection equipment, and a complete network architecture was constructed for information exchange and data transmission. The experimental results showed that the system had strong reliability and robustness, and the leakage detection algorithm has low false detection rate, which could effectively detect the abnormal temperature area. The image mosaic algorithm had high imaging quality and fast mosaic speed.