Accelerated Image Processing

Imaging systems used to be based on PCs. Expensive cameras connected with cables into an expensive frame grabber in a big computer box running Windows.

A new way has evolved in recent years: low cost embedded systems based on Linux, handling both the camera sensors and the image processing as well as the host/cloud communication, all this in small packages.

At BitSim we are doers. That means we handle the implementation part and get it to work: developing the hardware, translating an imaging design idea into algorithms, circuit boards, C/C++/Python code and accelerating functions in hardware based on powerful processing platforms.

We are focusing on two platforms:

– NXP’s IMX System on-chip CPUs: highly integrated ARM processors, GPUs, video encoders, and advanced neural network processing.
This is a stable industrial solution preferred by many customers.

– Xilinx FPGAs: Processing in  software and hardware,  using VHDL.
Programmable hardware inside the chip with integrated ARM processors cores or simpler “soft” CPUs. Programmable hardware to get the most out of your system: performance, low power, short latency etc.

Our own Python-based development platform SpiderPig is both embedded and at the same time identical to traditional inteactive PC-based environments (Python + Jupyter Notebook). Once you are ready with you algorithm based on OpenCV or machine learning you can convert it to use the same key hardware components on your customized circuit board. Same CPU. Same FPGA. Same I/O.