Vision based smart sorting machine

• Main Authors: Ng, Weng Seng, Ahmad Shahrizan, Abdul Ghani
• Format: Book Section
• Language: English; English; English
• Published: Universiti Malaysia Pahang 2018
• Subjects: TJ Mechanical engineering and machinery; TS Manufactures
• Online Access: http://umpir.ump.edu.my/id/eprint/23543/8/Vision%20based%20smart%20sorting%20machine%20CONF1.pdf; http://umpir.ump.edu.my/id/eprint/23543/14/2.%20Vision%20based%20smart%20sorting%20machine.pdf; http://umpir.ump.edu.my/id/eprint/23543/15/2.1%20Vision%20based%20smart%20sorting%20machine.pdf; http://umpir.ump.edu.my/id/eprint/23543/; https://link.springer.com/chapter/10.1007/978-981-13-8323-6_2

In this paper, a research on improved image processing method and a prototype of a vision based sorting machine have been developed to segregate objects based on color, shape and size. In today’s world, image processing is becoming popular technology and it grabs great attentions due to its capabilities of doing various applications in many field. The existing sorting system in industrial environment has to be improved by implementing the image processing method in the system. In some light industries, sorting process will be carried out by manually using human labour. However, this traditional method has brought some disadvantages such as human mistake, slow in work speed, inaccuracy and high cost due to the manpower. A vision based smart sorting machine is proposed to solve the aforementioned problems by segregating the workpieces based on their color, shape and size. It will be operated by a singleboard mini-computer called Raspberry Pi to perform the operation. In the proposed system, Raspberry Pi camera is used to capture the image/stream video of the incoming workpieces through the conveyor. The image/video stream of the incoming workpiece will be captured and implemented with pre- processing that consists of image enhancement to reduce the effect of non-uniform illumination which results from the surrounding llumination. To detect the color of the workpiece, the pre-enhanced image will be decomposed into its respective channels and the dominant color channel will be regarded as the object color. The result will be then matched with the database which is pre- installed in the raspberry storage through features matching method. The results from the features matching will turn on the servo motor and separates the workpieces’ color. For the purpose of shape segregation, the captured image will be first converted into black and white image before it is matched with the database based on certain coverage object properties. While for size segregation, the coverage object pixel area of the pre-processing image is extracted and matched with the databased in the system. Tested results indicate that vision based automatic segregation system improves the accuracy and efficiency of the works and thus the production rate of the industry