Counting rectangular objects on conveyors using machine vision
Counting the number of objects that are transported on a conveyor belt is frequently encountered in production facilities, airports or post offices. Although most of these tasks may usually be solved by using common photoelectric or inductive sensors, there are cases when objects have to be counted using more complex sensing systems based on machine vision. In this paper, an image-processing algorithm for segmenting, detecting and counting rectangular objects which are being transported on a conveyor belt is presented. The method is specifically designed to detect rectangular objects that can be partly occluded. The application is implemented using OpenCV/C++ library. Two different test scenarios are analyzed in the paper. Experimental results suggest that the proposed method has promising accuracy and it is applicable in real-world applications.