SCIENCE

Mercury Computer Systems announced its Imaging Toolkit, which delivers breakthrough capabilities for electro-optical, infrared (EO/IR), hyperspectral and radar imaging application development for the defense and commercial markets. The first to leverage the performance power of general purpose graphics processing units (GPGPUs), the Imaging Toolkit helps developers speed time-to-market and lower the overall total cost of development.

"Information is vital to scores of defense and commercial applications. Today, unmanned aerial vehicles (UAVs) with multiple sensors stay aloft for long periods of time performing persistent surveillance; manned vehicles perform operator offload functions such as scene and target tracking; and new applications as diverse as border security, fusion and exploitation, and natural resource management, use advanced sensors on remote platforms. These and similar vital programs collect incredibly large quantities of data that must be processed, analyzed and acted on virtually at once," said Brian Perry, vice president and general manager of Mercury's Services and Systems Integration (SSI) group.

Major challenges in processing and exploiting sensor information from data-rich applications include combining multiple imaging frames in real-time to create instantly actionable information for the people who need it. Mercury's new Imaging Toolkit, optimized for maximum performance on GPGPUs, helps customers address these challenges by providing an open framework to integrate and synchronize an array of sensors through the collection of highly efficient imaging functions. GPGPUs, because they are ideal for performing parallel processing/high-bandwidth imaging on large data sets, serve as the math co-processors for the embedded computing subsystems.

Although GPGPUs traditionally have been difficult to program, Mercury's Imaging Toolkit provides a layer of abstraction between the application and the GPGPU code, insulating developers from the GPGPU implementation details. The algorithms used in the Imaging Toolkit include imaging functions specifically designed to support high-bandwidth data streams from multiple sensors. These are optimized to take full advantage of parallel multicore processing while minimizing data latency, lowering risk, and allowing developers to focus on their application development.

A History of GPGPU Innovation

The Imaging Toolkit leverages Mercury's seven-year track record of bringing the tremendous processing advantages of GPGPUs to a wide range of defense and commercial applications. Mercury first implemented GPGPUs in 2004 as part of a medical imaging application and continues to deliver numerous GPGPU-based solutions, including a wide range of currently deployed, high-performance intelligence, surveillance and reconnaissance (ISR) subsystems.

The first of several customizable toolkits planned by Mercury, the Imaging Toolkit will be enhanced as technology advances require. Mercury's Services and Systems Integration (SSI) engineers, who perform the customization, engage closely with customers to develop the highest performance subsystems to meet stringent size, weight and power (SWaP) requirements and withstand a wide range of harsh environmental conditions. In addition to using Mercury hardware components, SSI can also work with third-party hardware, software and IP to deliver the optimal subsystem solution for the application.