Video-based broken filaments automatic detection and counting

Broken filaments is a common appearance defect of viscose filament surface, the number of broken filaments is a important quality index, the detection and counting of broken filaments is a important problem of quality inspection and control for viscose filament manufacturer. This paper considers this problem as random moving object detection and counting. First, viscose filament video is recorded, and is read into the MATLAB workspace; then a video image processing algorithm is proposed to process the video, and the target video stream with random broken filaments is derived; finally a method based on observation window is adopted to detect and count the broken filaments. Experimental results based on MATLAB show that this method can effectively detect and count the broken filaments, its accuracy rate can more than 90%. It turns out that the method has the potential to be used in the practical production.