Design And Fabrication of Color Sorting Machine Based on Computer Vision

Abstract

Manual sorting of objects by color can be exhausting and time-consuming, resulting in high labor costs and the potential for errors. At the same time, achieving accurate automation of this sorting process poses a challenging engineering task. This study presents a real-time method utilizing image processing of video frames collected by a camera and processed by an onboard Raspberry Pi computer. Unlike systems relying on color sensors or AI, this work achieves comparable accuracy using cost-effective, fast color identification with different colors' shades. An algorithm was specifically developed using the OpenCV package for this objective. The sorted objects are then transferred by a robotic gripper mechanism to their respective bins. The mechanical side will have a robotic arm and conveyor belt. The proposed design will be proven to be an effective and economical solution based on its capacity to pass tests set to evaluate its performance, where it exhibited a 90% accuracy under ideal operating conditions. The strategy being proposed is ideal for industrial applications that require color-sorting machinery, such as the fruit packaging and food processing sectors.