Industrial cameras are a key element of machine vision systems. Typical applications include automatic imaging-based inspection, process control, robot guidance, surveillance, microscopy, motion analysis, mapping, document digitizing, as well as medical imaging. Data collected by cameras are usually processed by a computer, with or without artificial intelligence.
Camera systems are based on a CMOS imager, a processing SoC or FPGA, and interfaces to transfer images to the "outside world." Depending on the application, imagers of different resolutions and frame rates are chosen. CMOS image sensors exist in different sizes. In general, the larger the sensor size, the better the dynamic range and signal-to-noise (SNR) ratio.
An SoC or FPGA processes the data collected by the image sensor. Typical processing includes creating usable still images or a video stream from the raw sensor data and compressing them for transmission. More advanced processing such as pattern recognition can be performed either in-camera, or in a central computer.
Various interfaces to connect cameras to other elements of a machine vision system exist. The most common are Ethernet, USB, GMSL (Analog Devices), V3-Link and FPD-Link (Texas Instruments). Wi-Fi or proprietary wireless interfaces can also be used.