Hyperspectral Imager With Real Time Detection and Identification For Small UAS

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

NAVAIR Public Release 11-017 Approved for public release; distribution is unlimited.

Hyperspectral Imager with Real Time Detection and Identification for Small UAS
Xiomas Technologies 1317 Skyway Drive Ypsilanti, MI 48197

Contact: John M. Green Phone: (734) 646-6535 Email: johngreen@xiomas.com Website: www.xiomas.com Command: NAVAIR Topic: N08-155 PROBLEM STATEMENT Current hyperspectral imaging system data rates exceed the bandwidth of small tactical Unmanned Aerial Systems (UAS). An automated analysis algorithm for tuning the hyperspectral output stream for optimum detection of specified target features in a given environment in near real-time and storage of the complete data set for post mission analysis is needed. Current systems ability to detect objects of interest such as Improvised Explosive Devices (IEDs) require improved attributes for use on small tactical UAS including reduced volume, weight, and higher spatial resolution. The hyperspectral data set require optimization for transmission across the existing data link, which currently handles 30 frames/sec video. Hyperspectral data collected is highly complex and require simplification for analysis in near real-time and post-mission. New technologies are needed to provide these capabilities including determining the optimal multi-spectral bands for rapid target detection, storing extensive mission data in its entirety, and parsing resultant data and metadata for transmission from the hyperspectral imager. 1

Advanced Ceramics Silver Fox shown with Xiomas Hypersepctral Imager. Silver Fox image is from www.acrtucson.com/UAV/silverfox/index

WHO CAN BENEFIT? This SBIR technology development is being sponsored by PMA-264 (Air AntiSubmarine Warfare (ASW) Systems) and PMA-290 (Maritime Patrol and Recon Aircraft). The technology is being designed for initial use and demonstration on a small, tactical UAS, the Silver Fox. The technology is applicable to other small UAS that require imaging systems designed to stay with the tight envelope constraints allowed
1

Navy SBIR 2008.2 - Topic N08-155, http://www.navysbir.com/n08_2/N082-155.htm

Xiomas Technologies, L.L.C.

Page 1 of 8

NAVAIR Public Release 11-017 Approved for public release; distribution is unlimited.

regarding size, weight and power (SWaP). Other areas and applications that could be of interest for the hyperspectral imager include but are not limited to: 1) Nuclear Non-Proliferation 2) Environment research and monitoring 3) Coral reef mapping 4) Oil spill mapping 6) Search and rescue 7) Crop management 8) Department of Homeland Security (DHS) for border control

BASELINE TECHNOLOGY Currently small UASs are equipped with electro-optical and infrared sensors, which provide good overall situational awareness but lack high performance target detection capabilities, and current high performance hyperspectral target detection systems are too large for use in small UASs. For example the Silver Fox currently carries the following Electro-Optical/Infrared (EO/IR) payload:

(Source: http://www.acrtucson.com/UAV/silverfox/UAV_SF_Brochure.pdf)

Hyperspectral imagers can be thought of as cameras which image hundreds of colors. Most people are familiar with black and white and color photography. Using black and white photography we can identify cars from trucks. We get additional information using a color camera, for instance, we can tell the difference between a blue car and a red car. While color photography uses three colors (red, green, and blue) hyperspectral imagers separate the light into hundreds of colors. This allows us to use computer based analysis to identify materials, which may look the same to the naked eye or using conventional photography. The science of hyperspectral material identification is well established and has been widely used in laboratories and in airborne sensors for decades.

Xiomas Technologies, L.L.C.

Page 2 of 8

NAVAIR Public Release 11-017 Approved for public release; distribution is unlimited.

The advancements we are pursuing with this project are to: 1) reduce the size of the device so that it can be operated in a small unmanned aircraft, and 2) to automate the complex analysis required to identify objects of particular interest so the system can be operated by soldiers in the field.

TECHNOLOGY DESCRIPTION The system fits within the size weight and power envelopes of the Silver Fox or similar UAS (approximately a 4.75" diameter by 12" long cylinder), weighs less than 8 pounds, is robust and provides a high probability of target detection and identification of targets of interest in desert, forest, and marine environments.

Xiomas Hyperspectral Imager Block Diagram with preliminary Optical Design The system uses a combination of Field Programmable Gate Arrays (FPGAs) embedded processors, and hyperspectral detection techniques, optimized by Xiomas, and an anamorphic hyperspectral imager built by Resonon, Inc (under contract to Xiomas). The anamorphic imager uses common fore-optics for both the ultra-violet/visible spectral imaging module (350-700 nm) and the infrared spectral imaging module (700-1,700 nm). This configuration results in very high performance in a small package.

Xiomas Technologies, L.L.C.

Page 3 of 8

NAVAIR Public Release 11-017 Approved for public release; distribution is unlimited.

The proposed system is composed of three main components: 1. Acquisition and Processing Unit (APU) 2. Hyperspectral Imager (HSI) 3. Real Time Automated Image Analyzer and Data Optimizer Acquisition and Processing Unit The APU is built around a small (about the size of a credit card) multipurpose processing module containing a high-performance Digital Signal Processor (DSP), FPGA, and General Purpose Processor (GPP) on a single chip. The APU includes enough data storage for an 8 hour mission, built in test capability for improved reliability, and a simple user interface. Hyperspectral Imager The hyperspectral sensor head is based on an anamorphic optical design, which provides fundamental advantages over current hyperspectral systems (such as the Offner) including: (1) The design eliminates the difficult imaging requirements along the long axis of the slit, which results in low optical aberrations and distortions and fewer optical components. (2) Because the anamorphic fore-optics are mirrors, they are achromatic and suitable for an instrument designed for a broad spectral range. (3) Perhaps most importantly, the anamorphic design partially decouples the spectral and spatial axes of the instrument. This results in engineering freedom not available with conventional designs that can be used to increase throughput. The bottom line result is high signal to noise ratio and low optical distortions. Signal to noise translates directly to target detection and identification. Our models indicate that a good rule of thumb is a Signal-to-Noise Ratio (SNR) greater than 100 is sufficient to identify targets which occupy as little as 1/3 of a pixel (the Xiomas system images a 4 inch pixel). Our analysis of the optical system predicts SNR well above 100 for all spectral bands as shown in the graphs on the following page:

Xiomas Technologies, L.L.C.

Page 4 of 8

NAVAIR Public Release 11-017 Approved for public release; distribution is unlimited.

SNR as a function of wavelength for the 700 1,700 nm module for a solar illuminated object with a reflectance of 0.3.

SNR as a function of wavelength for the 350 700 nm module for a solar illuminated object with a reflectance of 0.3.

Xiomas Technologies, L.L.C.

Page 5 of 8

NAVAIR Public Release 11-017 Approved for public release; distribution is unlimited.

Real Time Automated Image Analyzer and Data Optimizer The image analyzer is a flexible system that is capable of running a variety of algorithms, which Xiomas has optimized for use in a real-time environment. In real life the probability of detection of a sub pixel target is a rather complex matter involving; SNR, background, spectral difference compared to the background, solar illumination, etc. This system achieves a high probability of detecting small targets primarily through the high performance of the sensor optical design, and achieves a predictable performance by incorporating existing target detection techniques. During Phase II, we are implementing two techniques: Orthogonal Subspace Projection (OSP), and the Reed-Xiaoli Detector (RXD). OSP is a supervised technique, which involves (among other things), matching the collected real time imagery with an existing spectral library. This is a good technique for detecting 4 or 5 known materials. RXD is a more aggressive form of unsupervised data compression based on on-board anomaly detection. Hyperspectral anomaly detection is based on a generic information processing technique that produces a computational model for identifying cases in a data set that deviate from the norm. Each pixel is given a score that indicates how much it differs from the rest of the image. A threshold is then set and any pixel whose score exceeds that threshold is tagged as an anomaly. Table 1: Features, Advantages, and Benefits Feature Advantage
Anamorphic optical system High SNR: The Anamorphic system is expected to produce signal to noise performance of between 150 and 800 in a typical operational environment. Generally SNR above 100 is sufficient for target detection and identification. Can operate in a number of small unmanned aircraft such as ACR Silver Fox High spatial resolution improves the target detection results of hyperspectral image optimization algorithms Spectral response in the ultraviolet and visible range is important for imaging in the water while the longer wavelength are especially useful for camouflage detection and identification of specific objects and materials. Allows operation with minimal training

Benefit
Improved target detection and identification in a smaller package

Imaging system weighs less than 8 pounds High spatial resolution resulting in 4 inch ground spatial dimension (pixel size) Wide spectral range, 380 nm to 1.7 um

Flexible operation and ease of deployment Allows detection of smaller objects such as small IEDs

Allows identification of target(s) in diverse environments including but not limited to: shallow waters, deserts, jungle Reduction in training costs with reduced chance for human errors

Autonomous operation and automatic target detection

Xiomas Technologies, L.L.C.

Page 6 of 8

NAVAIR Public Release 11-017 Approved for public release; distribution is unlimited.

CURRENT STATE OF DEVELOPMENT


As of November 2010, we are at a TRL 3. Technology Development Milestones and Schedule:

Technical Requirements White paper Completed: 11-17-2008 Preliminary Design Review -- Analysis demonstrating feasibility of the hyperspectral Imager, the image acquisition and processing unit, and the target detection algorithm Completed: 6-11-2009 Critical Design Review Scheduled for 12-16-2010 Subsystem Development and Test including: o Hyperspectral Imager optical performance tests o Acquisition and Processing Unit data throughput tests o FPGA base Real Time Target Detection Tests with Simulated hyperspectral Scene Scheduled for 2-1-2011 System Functional Test and Flight tests Scheduled for 6-1-2011

REFERENCES Technical Point of Contact NAVAIR Acoustic Systems Division NAWCAD (301) 342-2094 Resonon, Inc. Rand Swanson, President Ph. (406) 586-3356 WHENTHETECHNOLOGYWILLBEREADYFORUSE The system will be at a TRL 7 at the end of the Phase II (winter of 2011) following flight tests and demonstrations. It is our intention to pursue additional funding from the Navy to advance the technology to TRL 9 and produce fully operational systems.

Xiomas Technologies, L.L.C.

Page 7 of 8

NAVAIR Public Release 11-017 Approved for public release; distribution is unlimited.

ABOUT THE COMPANY Xiomas is an early stage start up engaged in the development of high performance imaging systems for environmental research and monitoring, homeland security, and defense. Going forward we intend to develop, manufacture, and support, advanced imaging systems. Key Personnel and Bibliography of Directly Related Work John M. Green Xiomas founder and system engineer. Mr. Green has worked in the airborne imaging business for 20 years as a Product Engineer and Business Developer, mostly for Argon ST (formerly Daedalus Enterprises). Mr. Green has been involved in the development and operation of a number of instruments in use throughout the world including the MODIS Airborne Simulator (MAS) operated by NASA Ames Research Center, the MIVIS hyperspectral sensor operated by CNR in Italy, the Large Area Fast Spectrometer operated by the U.S. Marines, and the Airborne Multispectral Scanner operated by the U.S. DoD. Mr. Green has a Secret clearance and has been involved with a number of U.S. government programs. David B. Rein Xiomas founder, and Computer Engineer. Mr. Rein has 35 years of experience in the design and development of data acquisition and airborne imaging systems. His areas of expertise include; development of real time imbedded controllers, image processing, HW/SW interface programming, and DSP/FPGA software engineering. Mr. Rein received a BS EE from the University of Michigan in 1971. Brad Cook Xiomas Electrical Engineer More than 10 years related experience in camera/image/video acquisition and FPGA processing, system integration/debug of complex digital systems, and design/development of microprocessor/memory/communication circuitry and modules.

Xiomas Technologies, L.L.C.

Page 8 of 8

You might also like