A signature based approach for quality checking of biscuits

Quality checking is a major aspect in production. Fatigue and boredom are major parameters of human involved operating which causes for efficiency and productivity. Therefore Biscuit manufacturing industries have to overcome these situations throughout finding novel ways and methods for their human involved quality inspection systems. This paper proposes a computer vision based quality checking system for the biscuit manufacturing industry. A signature based quality inspecting system was introduced with a backlighting system to identify cracked biscuits which are coming out from a biscuit oven. A biscuit quality parameter value was introduced to the system, either can be programmed according to the quality requirement or an intelligent control approach could be introduced. Also this project addresses the possible noise additions in real industrial situations and the filtering of it. MATLAB Image Acquisition Tool Box and MATLAB Image Processing Tool Box with a 480K pixels web camera were used for the testing purposes and 73.33% accuracy was attained for various shapes of biscuits. The use of backlighting reduced the accuracy of identifying surface cracks by 10% for biscuits. This can be improved by front lighting or the use of strobe lighting which are application dependent. This methodology was tested under static conditions and can be alter and improve for moving biscuits on a conveyor. Also introducing an intelligent approach for the variation parameter will make this system more robust.