Design of Real-time Steel Bars Recognition System Based on Machine Vision

Currently, the bar counting is mainly completed by manpower, this method can make the workers fatigue easily, besides error counting, which can not guarantee the accurateness of packaging and match the high level of automatic equipment of steel rolling production line. To solve these problems, using machine vision and computer image processing technology, we design a real-time steel bars system. First we build up the hardware and software structure of system. Then we introduce the detailed working process, and design the detailed image processing algorithm. The results of experiments demonstrate the accuracy of bar counting in a single frame is up to 96% in this system, and its processing speed can meet the real-time requirements.