On-line longitudinal rip detection of conveyor belts based on machine vision

Abstract

Longitudinal rip of conveyor belts is a serious threat to safety production. Based on the machine vision technology, an algorithm used to find longitudinal rip of belts on-line from gray belt images directly is proposed. A gray image is first translated into a unidimensional vector. The unidimensional vector is further analyzed to obtain rip eigenfunctions. Then, faults of longitudinal rips are diagnosed by using the rip eigenfunction. The calculation of searching from the unidimensional vector is smaller than searching from the gray image. The validity of the proposed algorithm is testified by the testing results with some belt images.