Feasibility study of recognising apple orientations in Keyence vision system for industrial automation

Image processing on a smart camera starts at early vision, focused on enhancing a given feature in a scene, e.g., edge detection, and eventually ends up with image understanding and decision-making. The development of vision system spans a great variety of fields, and levels, from hardware to software, ranging from CCD up to CMOS integrated deep learning, therefore more and more smart camera applications are widely used in industrial automation.In order to make the apple packaging line automated by some robots, the key part is to recognise the real-time apple orientation on the conveyor by the turkey solution. In this thesis I will study the vision system and catch up with two ways to identify the orientation in the Keyence vision system, one according to the Profile Pattern tool, and the other according to Detect stain/flaw with contrast with background. Finally, I make some experiments to verify the feasibility of the two approaches.