Treffer: Optical Inspection System Using Real-Time, Embedded Digital Image Processing

Title:
Optical Inspection System Using Real-Time, Embedded Digital Image Processing
Contributors:
James Kang, Zhen Yu, Tim Lin
Publisher Information:
California State Polytechnic University, Pomona
Engineering
Electrical & Computer Engineering
Document Type:
Dissertation thesis
Language:
English
Accession Number:
edsbas.DD5F2AD9
Database:
BASE

Weitere Informationen

Automated optical inspection has become a much more common tool used in many manufacturing processes. Its benefits have helped production of goods become more efficient and cost effective while maintaining the desired levels of quality. Nevertheless, this technology is still a very high-end option that many smaller manufacturers are not able to get their hands on due to its complexity and cost. However, advancements in microcontroller technology and software development allow for the implementation of such a system at a much lower cost and level of complexity. This project aims to prove this point by proposing and implementing an automated optical inspection system that inspects the assembly of a particular Printed Circuit Board (PCB); inspection criteria include checking for the presence or absence of parts with the potential to check part alignment and functionality. The main hardware components for accomplishing this system consist of a custom-design, 3D-printed fixture that holds the camera and processor unit; this processing unit features an STM32H743II ARM Cortex M7 processor paired to a OV5640 image sensor capable of taking up to 2592x1944 (5MP) images. For simplicity purposes, the unit under inspection consists of a custom-design Light Emitting Diode (LED) matrix featuring an 8x8 array of WS2812B serial RGB LEDs. In terms of software development, this system takes advantage of the ARM processor being able to run the Micropython core; Micropython is a subset of Python ported over for use in embedded systems. This development platform allows for the use of many useful Python features as well as making it easier to perform cross-development on small parts of the project between a PC and the target system. The image processing implemented for this system takes a direct and mathematical approach using Morphological Operators such as Erosion and Dilation as well as edge detection algorithms. The results achieved through this approach, while not perfect or production-ready, are an acceptable foundation to build ...