Company

Signatur ITS is a problem solver for the ITS industry. Our customers are capable private and public companies, but they have lots of issues on their minds. Our sole focus is boosting performance of ITS systems by utilizing customers’ data better. Heard about “the real value is in the data''. That is what we do for our customers; help them get maximum value from their data.

Our company started life in 2011 as a spin-off from a R&D project on imaging and OCR technology. True to our legacy we continue to invest time and money into technical ingenuity and innovations, especially in imaging which still is our main technology area. Though we are not a large company, strong focus and a highly competent staff give us reason to claim leadership in our business niche.

Cameras are powerful sensors and images are terrific data. However, customers may have or need other types of sensors and data to reap maximum value from their ITS systems. Consequently, we have added top competence on some other key technologies as well; lasers being a good example.

Team

Geir H. Torp

CEO / Board Member

geir.h.torp@signatur-its.com

linkedin icon

Rune Kvisten

Project Manager

rune.kvisten@signatur-its.com

linkedin icon

Patrik G. Olsen

Software Developer

patrik@signatur-its.com

linkedin icon

Rune Lende

Software Developer

rune.lende@signatur-its.com

linkedin icon

Lars Aurdal

Research and Development

lars@findable.no

linkedin icon

Espen Noreng

Software Developer

espen.noreng@signatur-its.com

linkedin icon

Bjarne Olav Tveit

Project Engineer

bo_tveit@hotmail.com

linkedin icon

Christian Wilhelmsen

Hardware Engineer

christian.wilhelmsen@signatur-its.com

linkedin icon

Øyvind Nordvik

Chairman of the Board

oyvind@norict.com

linkedin icon

Anders Granquist

Board Member

anders@granquist.no

linkedin icon

Robert Sasak

Advisory Board Member

robert.sasak@gmail.com

linkedin icon

Erik Vasaasen

Advisory Board Member

erik.vasaasen@gmail.com

linkedin icon

Research

Matching images of the same vehicle; powerful and difficult

The image matching method developed and tested in the R&D project with the Norwegian Public Roads Administration (see References section) has much wider application than only within automatic control of average speed. Take for example tunnel safety; if you are able to match entry and exit images of the same vehicle, you know if that vehicle has left or still is inside the tunnel. Critical information in case of an emergency. Or in parking; matching entry and exit images of the same vehicle gives you parking time.

This seemingly simple task of matching images of the same vehicle at different times and places is much harder than one would expect. We may think that advanced software such as ANPR will master this task. In many instances that is true, but computer software operates at pixel level. And this makes ANPR vulnerable to reduced readability of images. We as human beings, however, grow up using and training our eyes as our primary “sensor”. We use the complete “picture” to understand and compare. This is a far more advanced “method” than most standard ANPR software use.

Superior matching rate opens up new areas of application

In the R&D project we have developed a method based on artificial neural networks (ANN) technology to compare and match entry and exit images of vehicles. This method utilizes details from the whole front of the vehicle, incl. the license plate, to do the matching. Though the project is not yet finalized, we see already that the method has much higher matching performance than ANPR. In practical use, e.g. matching vehicles in and out of a road tunnel, the number of possible matching candidates is limited. This makes it possible to achieve close to 100 % match rate, making the method suitable for areas which ANPR would not work due to an impractical high error rate.

Automatic anonymization according to GDPR

Part of the R&D project has been the development of a method for automatic anonymization of images according to GDPR regulations. Developing and testing ANN based methods require processing and storing of large volumes of images as training data. Without an automatic method the task of anonymization would have been daunting.