Czech algorithms from FIT aid digitalisation of Finnish timber industry

Automatic wood quality recognition and choosing of an appropriate manner of further processing. In brief, that is the plan of Tuomas Eerola from the Finnish Lappeenranta University of Technology who is currently visiting the Faculty of Information Technology of BUT and uses the algorithms and experience of his Czech colleagues in his research. The purpose of the DigiSaw project is to help modernising the timber industry and make the use of harvested wood as effective as possible.

At a glance, wood and computer technology do not have anything in common. Yet, Finnish researcher Tuomas Eerola aims to use the latest technology and finding in the area of image processing in wood processing. "The focus of the DigiSaw project is visual inspection of wood for timber product manufacturing. Thanks to automation, sawmill owners can cut boards in a manner that maximises their quality," described Eerola. This increases both the effectiveness of processing of felled trees and profits of sawmills. Gradual digitalisation and performance of checks of the wood has already increased profits by 24 per cent.

Inspection is carried out both on felled trees and subsequently on boards. "First, we record the trunks using a camera or a roentgen. Then we search for wood defects in the pictures. Those are for example knots, rot and many other things. A decision on how the wood is going to be cut and what will it be used for is then made based on the findings. After the wood is cut, it is inspected using standard cameras," clarified the Finnish researcher.

At the moment, Finnish sawmills are examining boards using two different devices. The goal of Tuomas Eerola and the DigiSaw project is to combine the data from these two devices. This will significantly increase the effectiveness and speed of wood quality evaluation. In his work, he also uses algorithms which were previously developed by experts from FIT BUT. "We have been co-operating with the Lappeenranta University for a long time. We exchange students and researchers and we create joint publications. That is one of the reasons why Tuomas Eerola decided to work on his project here. We have rich experience with detection of defects and detection of objects in a video. Furthermore, we can use neural networks for image processing," noted Pavel Zemčík, the Dean of the Faculty of Information Technology of BUT. It is thanks to some of the algorithms from FIT BUT the real-time assessment of the properties of wood is possible. "First, we had to show them tens of thousands of photographs of healthy wood and of wood with defects, or mark the defects for them if appropriate. Our algorithms then learned to recognise defects themselves," added Zemčík, who then continued by saying that the experts from FIT BUT use these algorithms for example in projects dealing with assessment of car detection and many other areas.

According to Pavel Zemčík, DigiSaw is not the first timber industry-related project the faculty participates in. "Even though we have so far not worked on other projects that dealt directly with cutting the wood, we co-operated with out Finnish colleagues on research related to further processing of wood. In particular, we have dealt with the production of pulp; we monitored production processes and used imaging methods to search for air bubbles in solutions used in production. That is because the quality of the given solution is assessed based on the size of these bubbles," said Zemčík.

Tuomas Eerola will be working on improving the effectiveness and bringing innovation to the timber industry in Brno until June; he will then present the results to his Finnish partners involved in the project. However, he does not know when the improved software will be put to use in practice. "The path leading from an idea to implementation in the industry is always fairly long," added the researcher, who also employs his knowledge of automatic image processing for example in a project studying the behaviour and movement of seals. "We have a rare species of seal in Finland that is found nowhere else in the world. They have unique patterns on their skin that work just like a fingerprint and thanks to these patterns, we can detect specific individuals on photographs sent to us by biologists. In effect, the biologists can easily monitor how seals move along the shore and so on," concluded Eerola.