A real-time cigarettes counting and loose ends detection algorithm

In industrial application of cigarette detection, lack of cigarettes and loose end problems are widely exist. The proposed work presents a vision-based cigarettes detection algorithm which allows lack of cigarettes and loose end problems being detected in real-time. Firstly, image segmentation and morphological operation are introduced to distinguish cigarettes region and background. Three-line model is then applied to locate each cigarette. After that, centers of each cigarette are found. At last, by counting the number of pixels in each obtained area, the likelihood of lack of cigarettes and loose end problems are given. The experiment results illustrate that proposed algorithm has high recognition rate in both lack of cigarettes and loose ends detection while it could operate in real-time.