Conferences and supporting programme
Software Development Strategies for Reusing Code in New Embedded Vision Applications
According to the Aspencore Embedded Markets Study 2017 by far most of the software developers (81%) reuse in-house invented code. This is not surprising as this code represents the core IP of an innovative enterprise. As more and more companies which are active in the field of computerized machine vision - which is still mostly based on a classic PC setup (industrial PC, discrete special cable, discrete specialized boxed camera) - are now start to recognize the benefits of an embedded approach (less power consumption, less space and most notably: significantly less costs). Transferring existing software code from a classic PC setup to an embedded target can pose a number of challenges: Different operating systems (i.e. standard Windows vs. hardware specific Linux), different processor architecture (i.e. x86 vs. ARM, MIPS, etc.) and different camera interfaces (i.e. GigE vs. MIPI CSI-2) and so on. Well defined standards which are able to abstract these differences can help tremendously to reuse existing code. With GenICam the industrial machine vision industry already established such standards years ago and it is still maintaining them in order to keep them up do date with new technologies. GenICam standardizes higher level protocols for camera configuration and image data transmission for lower level protocols (i.e. GigE Vison on Gigabit Ethernet, USB3 Vision on USB 3.0, etc.). In addition it provides standardized APIs for camera configuration and image acquisition. GenICam reference implementations exist for various operating systems and processor architectures. In addition there are camera vendor specific SDKs available which are based on GenICam technology and which add even more user convenience to the camera APIs. The broader the choice of the SDKs supported operating system, supported processor architectures, the number of supported camera interface technologies, the more flexibility is offered to the user to move from one technology to another and the easier it is to port existing code to the new target. The MIPI CSI-2 interface which is of particular interest for embedded vision applications has not been covered by the GenICam standard so far. Different camera vendors are already working on a GenICam-like abstraction layer on top of CSI-2 which in turn will make migration to this interface similar to other GenICam camera interfaces. It’s likely that CSI-2 will play an important role in the world of GenICam in the foreseeable future too.
--- Date: 27.02.2018 Time: 4:30 PM - 5:00 PM Location: Conference Counter NCC Ost