Processing system for an enhanced vision system

Abstract

Enhanced Vision Systems (EVS) combines imagery from multiple sensors, possibly running at different frame rates and pixel counts, on to a display. In the case of a Helmet Mounted Display (HMD), the user line of sight is continuously changing with the result that the sensor pixels rendered on the display are changing in real time. In an EVS, the various sensors provide overlapping fields of view which requires stitching imagery together to provide a seamless mosaic to the user. Further, different modality sensors may be present requiring the fusion of imagery from the sensors. All of this takes place in a dynamic flight environment where the aircraft (with fixed mounted sensors) is changing position and orientation while the users are independently changing their lines of sight. In order to provide well registered, seamless imagery, very low throughput latencies are required, while dealing with huge volumes of data. This provides both algorithmic and processing challenges which must be overcome to provide a suitable system. This paper discusses system architecture, efficient stitching and fusing algorithms, and hardware implementation issues.