Abstract: In view of problems of difficult detection and high risk in manual vision inspection of balancing tail rope failures of hoist, an on-line monitoring and early warning system of balancing tail rope of hoist based on machine vision was designed. The system uses CCD camera to real-timely collect on-site images of tail ropes of hoist, extracts characteristic parameters of the image by integration projection and Hu invariant moment, and adopts image feature classifier to do pattern recognition to get fault information and judge fault. Meanwhile, the system can remote monitor state of the tail rope of hoist by local area network and Ethernet, and can find out failures of the hoisting tail ropes, such as twisted rope, broken rope, the disproportional spacing and bending deformation of tail ropes, which solves problem of long detection cycle and fail to find fault of tail ropes timely.