Special: Striving to Protect America
Imaging Technologies Strive to Protect America
Aiding Homeland Security
The environment in this country presents an almost infinite number of potential targets for terrorists, and our Department of Homeland Security (DHS) has the intimidating task of protecting these targets.
The need for clear analysis of data for evaluating, strategizing and planning has never been more crucial. Engineers at NASA´s Kennedy Space Center (KSC) have developed a series of imaging- processing technologies that have potential applications in areas such as pattern recognition, remote sensing, monitoring, biometrics, optical character recognition and handwriting analysis.
Software providing an improved means of pattern recognition was developed at KSC. Pattern-recognition software is valuable in homeland security, as it allows one to comb through enormous sets of data looking for links and patterns, to investigate what may have already occurred, and, more importantly, to predict what may be planned to occur. KSC´s software is a tractable and empirically accurate algorithm, along with an integrated framework, that results in a visual process for pose invariant pattern recognition (PIPR). The algorithm and overall framework take advantage of hypotheses provided by a high-level visual process, thereby attempting to extract a region in an image based on these hypotheses. NASA´s PIPR technology operates independently of rotation or scale of the object. PIPR also requires no advance knowledge of the characteristics of images to be analyzed and provides an explicit indicator-of-match.
PIPR uses two reference points and two image points instead of one. The selection of two boundary points whose gradient angles are parallel determines a unique parallel gradient identifier angle. This parallel angle is the basis for PIPR´s rotational and scalar invariance. As a result, PIPR can simplify the computational process by testing only a single hypothesis rather than testing multiple hypotheses through the range of possible rotations and scales. Additional algorithms incorporated into PIPR quantify the level of match, which allows PIPR to identify objects even when there is not an exact match. The DHS´s image analysts can use the PIPR software to locate a predefined or known shape in images, such as a missile or rocket in aerial images of a terrorist´s training camp.
Several other KSC software-imaging technologies that may have applications in homeland security are based on "fuzzy reasoning." Fuzzy reasoning is an approach to computing based on "degrees of truth" rather than the usual "true or false" logic on which the modern computer is based. Fuzzy reasoning seems closer to the way a brain works and is used in making machines do a better, more precise job. For example, the thermostat in an oven controls the temperature by turning the oven on when it gets below the set temperature and off when it gets above the set temperature. A thermostat using fuzzy reasoning would be able to have the oven running all the time, partially on, making the oven stay at the desired temperature and run more efficiently.
The first technology relies on edge detection via fuzzy reasoning and has tremendous applications in pattern recognition and surveillance. In imaging, edges carry the most important information, and accurate edge detection is vital to performing advanced image processing and analysis. NASA´s fuzzy reasoning edge detection (FRED) system uses heuristics that mimic the capability of humans to approximate solutions, making it ideal for detecting edges in noisy, cluttered environments and unfamiliar objects.
Because it requires no advance knowledge of image characteristics, FRED has significant advantages over standard techniques. This fuzzy-reasoning approach to detecting image edges significantly supersedes current and widely used edge-detection methods such as Sobel and Prewit. An image of a CD was taken as a sample to compare methods; the fuzzy-reasoning approach shows tiny edge details not detected by the other methods. FRED uses a 3-by-3 window to scan the whole image and perform a heuristic analysis to find an optimal intensity gradient based on a 3-by-3 center pixel. FRED then generates a crisp center pixel value based on evaluation of a fuzzy membership function with respect to the optimal intensity gradient.
NASA currently uses FRED in two critical systems: The first system is used to identify and track foreign object debris (FOD) during space shuttle liftoff and is a key component of the current analysis in the investigation of the space shuttle Columbia explosion. The second system, the Cable and Line Inspection System (CLIM), is used to test the space shuttle´s emergency escape system. In homeland security this technology can be used for anomaly or defect detection, surveillance, visual-motion control, face recognition, object tracking, handwriting recognition and robotic vision.
The next technology based on fuzzy reasoning is a new fast-computational technique developed to find an optimal binary image threshold. NASA´s fuzzy reasoning adaptive thresholding (FRAT) system is ideal for binarizing noisy, cluttered or textured gray-scale images. FRAT is faster and more reliable than other current, highly dependable methods. FRAT can transform a poorly faded signature, a weathered document or surveillance tape of a license plate into a clearer, readable image.
FRAT defines an image as an array of fuzzy singletons corresponding to image pixels. With two classes, background and foreground, the membership function is built based on the average gray level of each class, which is computed using the gray level histogram as average weight factor.
By using an unrestricted range and a straightforward triangular-type membership function, FRAT takes advantage of a simple linear function as the basis for its entropy measure. The entropy measure is then used as a cost function for the selection of the optimal image threshold.
Security efforts have increased drastically since the terrorists attacked the United States, and they will continue to intensify. The strategic objectives of homeland security are to prevent attacks, reduce America´s vulnerability, and minimize damage and recover from current and future attacks. Although improved imaging technologies are important to NASA´s operations, they also can have a great impact on achieving the objectives of this homeland security strategy. The three imaging technologies described are available for licensing and further development by industry.
NASA Official: Janelle Turner • Web Design: Printing
& Design Office, NASA Headquarters • Credits