Digital video and high-resolution image analysis now occupies a large and growing portion of the embedded landscape and with each new application the CPU intensive signal processing burden escalates. Static and full motion electronic images of objects, people, vehicles, scenery, and documents are the raw materials for a wide range of digital image analysis applications such as machine vision, medical imaging, facial recognition, intelligent surveillance, robotics, and military radar analysis. In its most basic form, an imaging sensor captures snippets of spectral data and converts them into a digital representation useful to analysis software. However, the range of data is incredible: a linear bar code image only translates into a few bytes of information while a live, high-definition video image requires a continuous stream of data representing billions of pixels.
These image analysis applications bring a new level of complexity to embedded systems. High-performance, multicore processors along with real time, on-the-fly compression techniques are necessary to efficiently capture the high resolution images needed for analysis. To tackle these challenges, several board and systems manufacturers have recently announced new embedded video signal processing platforms based on the improved performance of 2nd-Generation Intel® Core™ processors. This new, long-life embedded architecture includes numerous graphics enhancements including an integral graphics processor for high definition hardware image decoding. The graphics section includes an array of parallel hardware execution units to accelerate encoding and decoding of high definition video. The Intel® 2nd Generation Core™ processors also include a new 256-bit instruction set called Intel® Advanced Vector Extensions (AVX), which is optimized for vector and scalar data sets such as those found in image and video signal processing applications. The graphics processor and CPU cores also feature Intel® Turbo Boost Technology, where clock frequencies can be increased for short periods to handle heavy workloads.
Mercury Computer Systems was an early adopter of this high performance architecture with their announcement of the Ensemble 6000 Series OpenVPX Intel® Core™ i7 Quad-Core Next Generation LDS6521 module (See figure 1). With high-end radar, electronic warfare, and image processing applications in mind, the LDS6521 combines the quadcore Intel® 2nd Generation Core™ i7 processor, an external FPGA for user-application functions, and high-bandwidth communication fabrics in a single 6U OpenVPX slot. The module is available in an air-cooled version or a conduction-cooled module that complies with the Ruggedized Enhanced Design Implementation standard or VPX-REDI for harsh environment embedded applications. Supporting multiple Intelligence Surveillance and Reconnaissance (ISR) applications, the LDS6521features Mercury’s POET (Protocol Offload Engine Technology) fabric interconnects for both Serial RapidIO and 10 Gigabit Ethernet. In addition to a wide range of built-in I/O ports, the module also provides two PMC/XMC mezzanine sites for additional I/O or control functions. Mercury’s MultiCore Plus (MCP) open software environment and MultiCore Scientific Algorithm Library (MCSAL) provides the LDS6521 with access to a variety of stacks, middleware, libraries, and software tools optimized for multicore processors.
Curtiss Wright Controls Embedded Computing has also developed an OpenVPX digital signal processing board based on the enhanced performance of the 2nd generation Intel® Core™ i7 processors. The CHAMP-AV8 6U delivers peak signal processing performance up to 269 GFLOPS by incorporating two of the quad-core Intel® Core™ i7-2715QE processors and the 256-bit AVX floating point instruction set (See figure 2). The module also offers expanded performance and bandwidth advantages with the new PCI Express to Serial RapidIO protocol conversion technology from Integrated Device Technology. With an on-board XMC site, 8 GB of flash, and up to 16 GB of SDRAM, the CHAMP-AV8 fits applications with demanding storage, data logging, and sensor processing requirements. The module is also supported with a suite of software including Wind River’s VxWorks and the Linux operating systems. Additional software support includes Inter-Processor Communications (IPC) and Curtiss Wright Controls Continuum Vector AVX-optimized signal processing library.
Along with reduced power, integrated graphics, and faster floating point performance, Intel® 2nd Generation Core™ processors deliver programmable media architecture plus hardware-based signal processing. These powerful graphics features provide the embedded platform designer new tools to match the requirements of next-generation image analysis applications. If you think that Intel® 2nd Generation Core™ architecture fits your next image analysis project please feel free to exchange information and questions with fellow followers of the Intel® Embedded Community. Also, there is more to come as I cover the features and benefits of using the Intel® Atom™ E6xx processor architecture in digital signage applications.
To view other community content on sensing and analytics, see “Sensing & Analytics - Top Picks”
OpenSystems Media®, by special arrangement with Intel® Embedded Alliance
Curtiss Wright Controls Embedded Computing is an Affiliate member of the by Intel® Embedded Alliance. Mercury Computer Systems is a General member of the Alliance.