ABSTRACT
High quality images of tunnel surfaces are necessary for precise visual judgment of abnormal parts as a replacement for human visual inspection to prevent any accidents against their deterioration. However, there is a trade-off relationship between the efficiency and visual precision. Because the amount of motion blur increases as both the resolution of the recorded image and the driving speed of the vehicle equipped with the camera increase, it becomes difficult to record sharp images that include detailed textures. We previously proposed a method to compensate for the motion blur resulting from the monitoring vehicle by introducing oscillating motion of the galvanometer mirror to offset the vehicle speed. However, because of the lack of responsiveness of a galvanometer mirror, the system did not correspond with the maximum traveling speed of 100 km/h permitted in a standard 7-m high tunnel along a highway in Japan. Accordingly, in this research we use a method capable of realizing precise path tracking to extend the corresponding frequency spectrum in PID control by a pre-emphasis technique, which is usually used in communication engineering. We calculate the pre-emphasis coefficient beforehand to follow a sine wave with zero gain, and amplify the input signal. Moreover, we propose a method to combine both a vehicle compensated for conventional motion blur and the pre-emphasis technique, and update the system for health monitoring in tunnels. As an experimental result, we confirmed significant improvement of image quality to compensate for motion blur for 0.2 mm black-and-white stripes and a crack with a width of 0.1 mm with motion-blur compensation at 120 km/h computationally. At the same time, the following result was achieved: compatibility between high accuracy and high speed; compatibility between the galvanometer mirror size and response speed; two times faster performance than previously; compliance with the requirements of a Japanese highway to prevent accidents.
