Automated visual inspection of wooden materials

Varnishing of wood materials is a delicate process. Defects on the varnish film affect the aesthetical appearance of the final product, which diminishes its commercial value or even can make the piece unacceptable. Quality control performed by human visual inspection is a slow, subjective and not repeatable process. For this reason, the application of an automated visual inspection system for detection and classification of defects is desirable.

General

The aesthetic appearance of manufactured products plays a decisive role in the furniture and wood industry. For this reason, varnishing is considered a critical process. Defects on the varnish film can make a piece, that could have already been treated intensively, unapt for commercialization. This implies not only a loss of resources, but also of productive time. Nowadays, the inspection of varnished surfaces is performed by human operators, providing slow, subjective, and not repeatable results. The use of an automated visual inspection system yields significant improvements in these aspects.

On the other side, varnished surfaces are considered non-collaborative surfaces and constitute a challenge for an automated inspection. This latter is due to the fact that defects on such surfaces can be imperceptible under certain illumination and observation conditions. Additionally, wood, as a nature product, presents variability within and between species. For these reasons, a single image of the varnished surface does not contain enough information to detect and classify defects. In this sense, approaches based on image series with variable illumination and observation direction are advantageous.

Project objective

The aim of this project is to develop methods based on illumination techniques and image fusion to automatically detect, identify and classify defects on varnished wood surfaces. These methods must provide a robust and reliable detection of varnishing defects on wood surfaces. Moreover, they must be independent of the wood texture and finally, they must be adequate for the integration in the industrial environment.