There is a crescent need to produce Printed Circuit Boards (PCB) in a customized and efficient way, therefore, there is an effort from the scientific and industrial community to improve image processing techniques for PCB inspection. The methods proposed at this chapter aim the formation of a system to inspect SMD (Surface Mounted Devices) components in a SSP (Small Series Production), ensuring a satisfactory production quality. This way, a 3-step inspection system is proposed, formed by image preprocessing, feature extraction and evaluation components, based on characteristics related to shape, positioning and histogram of the component. The inspection machine used in this project is inserted in a cooperation among machines context, in order to provide a fully autonomous factory, coordinated by a multi-agent system. Experimental obtained results show that the proposed inspection system is suitable for the case, reaching a success rate above 89% when using actual components.
TopIntroduction
Printed circuit board production is growing economically and technologically in an expressive and constant rate, where technologically is evidenced by a consistent advance of electronic devices functionalities, and economically by a 60 billion USD share in the world market, considering only electronics manufacture and final equipment assembly (IPC, Association Connecting Electronics Industries, 2014). Concomitantly with technological evolution comes electronics devices personalization, bringing a raise of SSP in the production systems scenario, as a result of the possibility to produce in multiple perspective and manners (Hitomi, 1996). To produce a variety of products in a short time space production, a flexible manufacture process with high reliability is demanded, because each product has a high aggregated cost. To minimize the potential failures during production, inspections systems are used to identify these failures as soon as possible.
One approach of inspection is the automatic optical inspection (AOI), which is capable to detect deficiencies in a non-destructive and precise manner, in this case, detect failures in component placement, such as misalignment, shifting, wrong type or missing component.
This work is addressed to inspect SMD without considering any serigraphy marking. The inspection position in the production line is located after insertion of all components and before the welding process, hence we don't have the presence of solder paste in the PCB.
- •
Image Processing Pipeline: An image processing pipeline that permits an adequate visual automatic inspection of SMD components in a SSP context.
- •
Image Recognition Pipeline: An image recognition pipeline, based on artificial intelligence techniques, to identify the status of a component.
- •
An Inspection System Agent: An agent that represents the inspection machine in the multiagent environment, and the communication system between agent and machine.