Vision Components has developed an FPGA-based accelerator for edge pre-processing of image data in embedded vision projects with MIPI camera modules. The hardware accelerator with multiple MIPI-CSI-2 inputs and outputs enables complex image processing and image data analysis.
The compact board can merge data from several MIPI cameras. It manages complex algorithms and computing operations thanks to the powerful, freely programmable FPGA. The accelerator will be available early in 2022, initially with a completely open FPGA for customer programming and demo applications.
In a second phase, Vision Components will launch its own FPGA designs for specific applications such as color conversion, 1D barcode identification, and epipolar correction. In addition, there are plans to enable the electronics for AI acceleration. Vision developers can integrate the accelerator into their future electronics designs – just as easily as a conventional MIPI camera module. The accelerator transfers image pre-processing results to a higher-level CPU via the MIPI channel. This allows users to perform even very complex tasks by means of in-depth pre-processing in the dedicated FPGA module, while they also benefit from the flexible selection of a highly efficient embedded system.
Vision Components is also adding a variety of new MIPI camera modules to its existing range. The manufacturer has integrated a time-of-flight sensor – the Melexis MLX75027 – into a 50 mm x 50 mm board with a MIPI-CSI-2 interface and lighting. The sensor’s DepthSense technology with pixel sizes of only 10 μm x 10 μm enables high-resolution, high-contrast 3D imaging. The new ToF MIPI module will be available to order later in 2021, as will four additional MIPI camera boards with Sony Pregius S sensors – the newly integrated high-end sensors feature global shutters and resolutions up to 12 megapixels.
Another addition to Vision Components’ MIPI portfolio to follow is a line of MIPI camera modules for imaging in the short-wave infrared (SWIR) spectrum. Different sensor technologies will enable especially inexpensive SWIR applications.
- Fourth-generation global shutter explained, and why embedded image sensors need better performance metrics
- Global standardization efforts foster machine vision adoption
- New ams image sensors for high throughput industrial vision
- How MIPI Alliance specs enable the IIoT