A large number of different automatic number-plate recognition (ANPR) applications such as parking systems and toll control, requires that the system can be adapted in a short time and at low cost. Wahtari has developed the neural camera that is based on a flexible modular principle and equipped with Allied Vision’s Alvium 1800 C cameras.
Wahtari has developed solutions for a wide variety of tasks. Depending on the application’s scope and the requirements, end users can create a product tailored to their needs based on Wahtari’s open architecture.
For use in license plate recognition, the system must meet the highest requirements in terms of precision and speed. With a large number of different applications such as parking systems and toll control, it is essential that the system can be developed in a short time and at low cost.
Open system architecture
The neural camera nCam developed by Wahtari is based on a flexible modular principle. It consists of the basic elements of an image sensor in the form of an Alvium CSI-2 camera, a carrier board, a lens, an image processing controller, a network interface, an AI accelerator, and an x68_46 Intel processor. Up to three Intel Movidius Myriad X VPUs neural processing units are dedicated exclusively to accelerate sophisticated deep learning inferencing. With its high-performance industrial image sensors ranging from 0.5 to 20.4 megapixel resolution, Alvium cameras fit perfectly into the system’s overall concept.
As an open platform, nCam is used in many areas of intelligent traffic solutions (ITS), like license plate recognition (ANPR – automatic number plate recognition). Typical ANPR applications are vehicle access control e.g., via black & white lists, automatic billing of toll and parking tickets, as well as free-flow applications in multi-level parking garages or other constellations where RFID transponders are unsuitable.
Flexibility through modularity
Wahtari’s modular camera design is developed exactly for versatile requirements. The large sensor selection of the Alvium series opens up a wide range of applications for the nCam. Frame rates of up to 289fps or a resolution of up to 20.2MP are possible, while LEDs with white or IR light ensure a visibility range of up to 50m even at night. Depending on the task such as license plate recognition, the system can now be tailored exactly to individual requirements.
The overall system is no larger than 355.00mmx130.00mmx210.00mm. Packaged in an IP67 aluminum housing with a sophisticated passive cooling system, the camera reliably delivers results in a matter of milliseconds even under the harshest conditions (-40°C to 55°C/ -40°F to 131°F). Optionally, it can be equipped with a sunshield.
The ANPR system demonstrates its particularly high efficiency gains in scenarios where a large number of unknown vehicles need to be identified quickly and without errors. The number of incorrectly recognized license plates drops to almost zero thanks to deep learning. The collected data provides an ideal basis for drawing conclusions about the utilization of road or parking spaces, analyzing drivers’ usage behavior, and defining appropriate optimization measures.