The provider of software for industrial image processing has, according to its own information, technologically proven that its software can collect and process images from MIPI (Mobile Industry Processor Interface) camera modules as standard. Existing Halcon interfaces such as Video4Linux, GenTL or even direct memory access (shared memory) can be used for this purpose. To this end, MVTec's hardware interface experts successfully tested the VC MIPI OV 9281 camera module from Vision Components in conjunction with Raspberry Pi 3 and 4 with Halcon 19.05. With the same Halcon version, the MIPI IMX290 sensor module from The Imaging Source was also successfully tested via FPD-Link III (up to 15 m cable length) on an NVIDIA Jetson Nano. This proves for the first time the full compatibility of the image processing software with the camera modules specially developed for the embedded sector, which meet high industrial requirements and focus on high-volume applications.

Hardware manufacturers such as Nvidia, NXP or the Raspberry Pi Foundation already provide dedicated interfaces to MIPI as standard. The transmission standard allows various image sensors to be seamlessly integrated. This means that digital image data can in some cases be captured by embedded devices without any additional CPU load. This means that the full performance of the processors can be used for vision applications. Users also benefit from a maximum transfer rate of up to 17.1 GBit/s. With these advantages, MIPI also paves the way for completely new business areas outside of industrial image processing, not least due to the lower costs of embedded devices. These include, for example, medical technology, the agricultural industry and the logistics and retail sectors. New applications are now possible that could not be realized with embedded devices in the past due to unsuitable sensor modules.

Halcon features for all sectors

With this hardware configuration, MVTec will also support all Halcon core technologies such as matching, identification and measurement, as well as the latest features such as executable deep learning inference on Arm CPUs. This results in benefits for all industries. Joint demos are also planned with Vision Components and The Imaging Source at Embedded Vision Europe on October 24 and 25 in Stuttgart.