Design team members: Gregory Niestrawski
Supervisors: Dr. Alex Wong
Background
Fabrication processes are never perfect. When manufacturing electronic circuits, tiny particles of dust or debris, imperfect chemical etching, and uneven deposition of solder can radically change the intended electrical characteristics of a circuit. These defects are not always easily detected, and can prematurely end the life of products such as appliances, audio equipment, or computer components. Manufacturers often design expensive, specialized hardware to detect defects in electronic circuits before they are sold.
Project description
Manufacturing cost (and hopefully the price of our electronics!) could be dramatically reduced if computers could be reliably used to automatically detect defects. Computer vision (image processing) can be used to analyze digital photographs of circuits and find the defects. Algorithms have been designed in the past, but assumptions made the techniques impractical for most applications. This project attempts to design a robust algorithm which can be adjusted and applied to any circuit, using inexpensive imaging equipment.
Design methodology
- Determine what characteristics defects have
- Teach a computer to recognize defects mathematically based on those characteristics
- Photograph the circuit which may contain defects
- Fix the photograph to make the defects more visible
- Use computer vision to find the defects in the image!