Multi Video Processing on Embedded Platform

while we have applications available for processing still images on embedded platforms, there are none available for video processing, that too in real time!! Video processing requires good amount of memory as well as processing power. With IoT is set to become a reality soon, it is envisaged that video processing would also become a basic requirement for embedded systems. Coming to the aid of embedded systems designers and developers nowadays is the availability of powerful 32 / 64 bit MCUs with on-chip floating point and DSP support as well as significant amount of RAM memory. Further, a number of OS with small foot print yet with required drivers for interfacing different types of cameras and displays and also optimized libraries are available in the form of open source for video processing embedded applications development. This papers deals with the implementation of video processing functionality for embedded systems. Various ARM processor based boards and flavors of Linux were evaluated and, Raspberry Pi B+ board running the Raspbian OS was chosen as the platform for the implementation. A 3.2 TFT touch screen was identified as the display. The software application implemented incorporates an algorithm for processing the real-time video stream captured from a camera that was interfaced to the embedded platform. The work involved study with various resolutions, frame rates and angles for rotation as well as compression of the video frames so as to analyze capabilities of the 700MHz ARM application profile processor of the platform.