Smart Machine Vision Camera

Intelligent cameras not only record images, but also process them to recognize important image information themselves.

Cameras that can be easily configured, programmed and expanded have a wide range of applications. Thanks to their high level of adaptability, end users can use such a camera in various areas like in bottle filling lines for checking the level and label of the bottle. Further application examples are the fast sorting of wooden boards in a large sawmill on the basis of quality characteristics such as tolerable branch size, or the recognition of car numbers in traffic monitoring. In all these applications, the images are taken by the camera and processed thereon. Thus, the sorting system can be controlled directly and no additional computers and network infrastructure are needed for the image analysis. This not only simplifies installation but can also reduce product costs.

For the real-time-capable control of the image sensor and the preprocessing of the read-out image data digital circuits in the form of FPGAs are used. For example, they ensure that all products are checked in time and no unwanted latencies arise in production plants with high conveyor speeds. A processor or DSP (Digital Signal Processor) not only takes over the general control of the camera and controls the components involved, but also allows the implementation of user-specific functionality. The provision of an application programming interface (API) facilitates the camera operation and image management for the programmer. The support of popular software libraries like OpenCV favors the development of own applications further.

In order not to lose any image during processing, industrial cameras must support very high frame rates of several hundred frames per second and be able to process the resulting data in real time. A delayed image evaluation, missing an event or a delayed system control can have fatal consequences and lead to loss of production or even personal injury. In order to guarantee the required real-time behavior, sophisticated architecture concepts and implementations are required.

Application-specific lighting conditions and image quality requirements require the use of different image sensors. For this, the development and maintenance of cameras can be supported and facilitated by a modular and extensible concept for the integration of various CMOS and CCD image sensors. The unified control of the camera also simplifies the programming for the end user.