Automatic optical inspection (AOI) systems for printed circuit boards are used within the industry during the manufacturing process. The primary purpose of such systems is to detect irregularities on printed circuit boards. They contribute to improving the quality, reliability, and efficiency of printed circuit board production.
In this master's thesis, the possibility of simplifying and reducing the cost of such a system is explored, as these types of systems can be quite expensive.
Automatic optical inspection (AOI) systems for printed circuit boards in our master’s thesis was developed on the Raspberry Pi 4 platform. In order to capture images in a consistent environment, an illumination unit was created. The captured image of the printed circuit board is rotated using affine transformations so that the edges of the circuit board are parallel to the edges of the image. The rotated image is then cropped, leaving only the printed circuit board without background in the image. The application is divided into two parts. The first one emerges when the circuit contains fiducial markers and the second one when it does not. If the circuit doesn't have markers, the origin point of the coordinate system is determined immediately, and the edges on the image are detected. However, if the circuit contains markers, they are first located using a template matching method, and their positions establish the coordinate origin. After locating the edges through a subpixel accuracy algorithm, these edges are filtered to determine which components they belong to. In the final phase, based on the filtered edges, it is determined whether a component is correctly attached to the printed circuit board or not.
The system is capable of detecting errors in resistors, coils, and capacitors. Errors that such a system can detect include components that are too displaced from the center, missing components, and inverted components.
|
