Video-based roller abnormality detection

In view of the failure of the belt conveyor roller during operation, this article conducts non-contact detection on the roller through video analysis of the roller, and finds the abnormally operating roller. Firstly, use image texture extraction to extract and quantify the movement characteristics of the roller in the spatial domain into a series of time signal digital sequences, and then perform fast Fourier transform (FFT) on the signal digital sequence to obtain the rotating roller frequency to establish the relationship H(f) between the linear velocity of the roller surface and the rotation frequency of the roller, and finally the abnormality of the roller is tested according to H(f). Video-based roller abnormality detection effectively reduces the cost of manual roller inspection and improves the efficiency of inspection. Compared with the identification of roller abnormalities based on sound signals, video-based roller abnormality detection can locate specific abnormalities Roller.