.::APHESA SPRL - imaging, electronics and software consulting::.

Camera architectures

During our preliminary project meetings with customers, we are usually asked about the general architecture of a camera, so let's explain this in a short blog post.

Of course the camera starts with an image sensor and ends with a computer interface, that's the most common concept and we will restrict the discussion to this one. Between the image sensor and the computer interface, camera management or image processing has to take place. There is no camera that doesn't have a bit of image processing, at least to get a good image out of an image sensor, even though this is not actually machine vision image processing.

There are several architectures for that control, management and image processing.

  • FPGA without external memory - used when the output interface does not require data buffering and the processing does not require random access to the memory or the use of previous frames. We work with Xilinx FPGAs.
  • FPGA with external memory - used when the output interface requires data buffering and/or the processing requires random access to the memory or the use of previous frames. We work with Xilinx FPGAs.
  • DSP or processor, usually with external memory - for cost reasons or because of some complex image processing that requires software libraries. We work with real DSP like BlackFin or advanced microcontrollers like i.MX.
  • FPGA and DSP or processor, usually with external memory - in some cases we need to have access to the best of both worlds and therefore a DSP or processor is used for complex machine vision image processing tasks and the FPGA is used as computer interface, sensor interface and performs some hard coded image processing functions. In this approach, hybrid devices like Zynq offer a powerful solution.

Besides the main path, there are many other devices on board. They are used to store information like serial number, configuration or calibration. They can be temperature sensors, shock sensors, voltage and current sensors, LED drivers or other secondary functions. All these devices are hooked to a slow internal camera bus and connected to the primary device (FPGA, DSP or microcontroller). There are also many devices that require direct control without a common bus, for example many of the controlled power supplies that are required for some complex image sensors.

Most of the time, the FPGA contains a soft core microcontroller to perform all communication and slow speed control and calculation tasks.

More information about this is provided in our Imaging Technology courses.


About us

Aphesa develops custom cameras and custom electronics including FPGA code and embedded software. We also provide EMVA1288 test equipment and test services as well as consulting and training in machine vision and imaging technologies. Aphesa works in several markets including industrial, medical, oil&gas and security.