Real-time Image Processing

Industrial image processing and pattern recognition becomes further established as a key enabler technology in producing companies. Quality assurance via optical measurement strategies, machine conditioning and product analysis, as well as man-machine-interaction are some of the main issues in this area. The institute is working on interdisciplinary approaches based on technology, biology and perceptual psychology in the area of real-time image processing. The main focus lies on image processing methods on the one hand and algorithmic and implementation aspects on the other.
The target of the industrial image processing is to process image information from automation systems with regard to the process real-time, stability and limitation of resources. In the sense of a holistic approach of industrial systems image data as well as expert knowledge are consulted as information sources. Particularly, the description, the modeling and the design of effectively implementable algorithms for resource-limited microelectronic circuits like FPGAs as well as FPGA- and GPU-based systems are in the foreground of the research.

The application focus is currently divided into two areas: On the one hand, image processing concepts for the automation are processed. On the other hand, we are dealing with the authentication and inspection of currencies. This field of activity which is at the first glance quite diversified is at the same time the strength of the real-time image processing competence scope because it combines “the best of two worlds“. Therefore, interdisciplinary approaches using technical as well as human-oriented concepts can be applied. Product and material inspection topics are incorporated in the competence scope by automation. Human-perceptive image processing concepts are mainly driven by the field of authentication and inspection of currencies.
For instance, non-visible coding of surfaces for the authentication of machine parts in automation systems are mentioned here. The research and subsequent technical realisation in the application is only possible by the knowledge of the human visual system’s strengths and weaknesses.
On the other hand, the know-how of the exposed capacities of the visual system leads to pattern recognition and classification algorithms which are used in cameras in the area of product inspection. Besides the interdisciplinary image processing aspects, resource-efficient real-time algorithms for decentralized Smart Network Cameras and especially Mobile Devices such as Smartphones are in the focus of applications. These facts result in powerful “intelligent” systems for image processing tasks.
The inIT competence area of real-time image processing is one of the initiators of the network “industrial image processing OWL”, an initiative which sets itself the target to promote industrial image processing under interdisciplinary aspects for automation systems.

On 19 November 2014 the 4th annual colloquium on “Image processing in automation” (BVAu 2014) took place at the CENTRE INDUSTRIAL IT in Lemgo. The colloquium is held on an annually-alternating basis in Lemgo and Paderborn.
The colloquium sees itself as a forum for science and industry which covers an essential research area of automation since industrial image processing and pattern recognition “is a key technology for future products and at the same time the basis of ‘intelligent’ quality management systems.”
Furthermore, the team members published several peer-reviewed papers and submitted some patent applications together with partner companies.

 

Optical Document Security

A special branch of application is covered by Optical Document Security (ODS) which is, based on its conception, a notable field in applied research. In this field inIT collaborates with international institutions as Central Banks as well as with renowed companies.
The research focus is consequently oriented towards the holistic approach in the area of banknote and document security. This orientation is essential for the international cash handling cycle and personalised security. In particular, we consider the chain of the entire banknote lifecycle: production and quality management of banknotes, authentication and wear-and-tear as well as security in the area of Automated Teller Machines (ATMs).
Additionally, concepts and application-orientated approaches for brand label protection as well as document security are in the focus of research. Our strategies are based on the fields of competence in real-time image processing, sensor/information fusion and pattern recognition.

Within the scope of several research projects in the field of Optical Document Security. inIT has been working for considerable time on a procedure for authentication of banknotes. In 2014 remarkable success was reached in the project “Sound-of-Intaglio – Banknote Authentication with Printing Processes”.
Based on “Sound-of-Intaglio” a new approach for robust image capture with mobile devices was presented at Optical Document Security – The Conference on Optical Security and Counterfeit Deterrence IV in San Francisco. Furthermore, a new approach for impartial quality measures in Intaglio print was proposed. The approach will be implemented in a real-world-scenario in 2015 to prove its production efficiency in a production line. An additional work topic which was defined was the realization of an authentication and sorting system for banknotes which is able to handle up to 60 banknotes per minute. The system is used for lab tests and counterfeit detection analysis at external partners.
The ODS-Team has been frequently appreciated by the media regarding their expertise about banknotes, because in 2014 the new Europa-series of the 10 EURO banknote was introduced.
New methods and approaches for anomaly detection with information fusion concepts for anti-skimming on ATMs have been investigated in the framework of the leading-edge cluster “it’s OWL”. Additionally, material authentication approaches for ATMs were developed and conceptually realised. Moreover, research work is carried out on multidimensional vibration analysis for ATM-components.

To realise research tasks in the area of banknote printing machines, in particular anticipatory multi-sensor fusion-based fault analysis and condition monitoring, the roller demonstrator was integrated in the Lemgo Smart Factory and meanwhile has been expanded considerably.

Furthermore, the team members published several peer-reviewed papers, carried out keynotes and submitted some patent applications together with partner companies.

Professor

Prof. Dr. Volker Lohweg
E-Mail: volker.lohweg(at)hs-owl.de
Phone: +49 (0) 5261 - 702 2408
Fax: +49 (0) 5261 - 702 2409