Enhancing target tracking in critical applicationsStory
April 16, 2010
By using the latest generation of low-power, mixed media processing devices developed for the 3G smartphone market, an embedded tracker can be optimized for both performance and a platform?s vital Size, Weight, and Power (SWaP) characteristics.
Electro-Optical (EO) sensors continue to evolve in performance and resolution, requiring more signal processing power for better target discrimination, lower error rates, and improved accuracy. Automatic target tracking enables the sensor subsystem to identify and track multiple targets through clutter, weather, atmospheric disturbances, and countermeasures designed to obscure the target or seduce the sensor. At the platform level such as a combat aircraft, armored vehicle, or helicopter, it is often proposed that target tracking could be performed as a software task by an integrated central computing resource, receiving raw sensor data from an array of sensor types. Compared to a dedicated, embeddable hardware solution, this architecture requires considerable general-purpose processing power. However, by using the latest generation of low-power, mixed media processing devices developed for the 3G smartphone market, an embedded tracker can be optimized for both performance and a platform’s vital Size, Weight, and Power (SWaP) characteristics.
EO sensors are used extensively in ground-to-air, air-to-air, and air-to-ground fire control systems, surveillance, security, and perimeter defense systems. Sensor images are processed by a tracking system to identify and follow objects such as aircraft, vehicles, or personnel that are moving within a set of prescribed characteristics for those particular types of objects. Automatic tracking is often a requirement as it provides low error rates and consistent performance and does not suffer the same fatigue or stress as a man-in-the-loop. A typical tracker has three functional stages:
- Preprocessing the incoming video stream at the pixel level for contrast and statistically based enhancement
- Image analysis to identify objects of interest within defined areas or windows
- Analysis of objects’ movement over a number of frames to identify them as tracks, which are then followed using predictive algorithms to confirm their progress.
The complexity of the tracking function varies considerably from a fixed-position tracker for a single target fire control system, to a tracker on a highly maneuverable airborne platform such as an Unmanned Aerial Vehicle (UAV) or helicopter, required to track multiple ground-based moving targets. Many tracking applications, such as ground attack aircraft or helicopters, are closed-loop systems whereby the sensor and/or a weapons system must follow a designated target track. This requires rapid loop response time and hence dedicated high-performance processing to maintain lock on the target if the sensor platform and target are both moving.
3G mobile telephony provides key technology source
Sensor technology continues to evolve rapidly, requiring a comparable increase in signal processing power as pixel count, frame rates, and analog-to-digital conversion resolutions increase. Recent higher-resolution sensors transmit video in digital formats, including CameraLink, DVI, HD-SDI, or GbE Vision. The raw digital video data might contain a large amount of noise or clutter or have poor contrast. This often applies across the whole image, so a highly parallel processing device such as an FPGA is well suited to filtering the content, leaving more clearly distinguishable target data for detailed analysis.
Sophisticated tracking and classification algorithms applied to the reduced data set require the greater processing capability of a DSP device. GE Intelligent Platforms’ ADEPT 5000 high-performance, embeddable tracker combines these key FPGA, DSP, and 3G smartphone computing technologies to create a rugged, core tracking module small enough to be mounted on a PC/104-Plus sized (3.6 x 3.8 inches, 90 x 96 mm) I/O base card (Figure 1). These tracking modules can be attached to many open-standards I/O base card formats such as VPX, VMEbus, PC/104-Plus, PCI, or PCI Express. They enable construction of a complete embeddable tracking subsystem that could be colocated with the sensor or embedded as part of the platform’s mission computing or display processing resources.
Figure 1: The ADEPT 5000 high-performance tracker from GE Intelligent Platforms
Enhanced sensor platforms meet today’s demands
Ongoing budget concerns continue to restrict new platform development, favoring instead the enhancement and life extension of existing platforms. New or upgraded EO sensors are planned to provide these platforms with vital new operational performance. Advances in tracker capability are helping to hone this edge with the successful infusion of cross-market technology; this results in a complete dedicated tracking subsystem occupying a significantly smaller footprint than a comparable general-purpose embedded computing module, while consuming less than one-third of the power.
To learn more, e-mail Duncan at [email protected]