The past ten years have seen an explosion in the adoption of machine vision and image processing. Once considered too costly to be economically deployable at scale, the decreasing cost of sensor systems and an increase in compute power to process the video has resulted in the broad adoption of vision systems.
Two modern trends threaten to stretch the capabilities of image processing sub systems to their breaking point. The first is the availability of ever higher resolution sensors that are capable of outputting video at higher frame rates resulting in much higher data rates. The second is the proliferation of artificial intelligence and machine learning, which enables computer systems to extract more intelligence from the video, although at the cost of significantly higher processing requirements.
Only the very highest volume applications can support the considerable investment of custom designed video processing SoCs. For other applications, salvation is found with programmable logic. Efinix® FPGAs were designed with video processing applications in mind. They bring inherent parallelism and the possibility to design efficient pipelined architectures. They have embedded MIPI interfaces for natural connection to video sensors and an infinite combination of artificially intelligent processing systems can be instantiated in their programmable fabric. With low power consumption and cost effective silicon efficiency, Efinix Trion® and Titanium FPGAs have been widely deployed in a broad range of video processing and surveillance applications where they provide image enhancement, feature extraction and artificially intelligent analysis of the video data.