Paper Title
Implementation of Computer Vision based Industrial Automation Systems and Performance Evaluation

There is an ascending importance in automating the manufacturing, verification and control application in industries. The huge benefit of automation is that, it reduces labor, to save energy, to save material, to improve accuracy, precision and quality. The proposed research work is to resolve two issues concerned to industrial safety and energy losses. The first one is “Vision based Industrial fire detection and suppression” and second one is“Vision basednon-magnetic object detection in steel industry”. In the present developing world, accidental fires has become a big threat for large and medium scale industries involved in production or processing. Automated fire suppression system plays a very significant role in Onsite Emergency System (OES) as it can prevent accidents and losses. A rule based generic collective model for the classification of fire pixels is proposed with single camera view with multiple fire suppression which sprays chemical through control valves. Neuro- Fuzzy algorithm is used to identify the exact location of fire and its suppression in the image frame. Finally SCADA system has been explored to operate the fire suppression control valve. To verify the performance of the algorithm 150 videos have been given as input with various fire and non-fire sequences. The results are encouraging with an accuracy of 98% and 95% in the case of fire detection and suppression with the computational time of not more than 1900 milliseconds. Vision Based Non-Magnetic Object detection on moving conveyor in steel industry is developed by considering single static background. Establishing background and subtraction of background from continuous video image sequences forms the basis. Detection of non-magnetic materials which are moving with raw materials and taking an immediate action at the same stage of material handling system will avoid the breakdowns and power wastage in the industry. Proposed system has resulted 96% accuracy in non-magnetic object detection with computational time of not more than 1300 milliseconds. Keywords - Fire Rules, Neuro-Fuzzy Algorithms, Motion analysis, Differential Methods, SCADA.