Abacus®

Hyperspectral Image Processing
Aerial views of the earth's surface have intrigued people since the first experiments in aerial photography occurred from US Civil War surveillance balloons. Since then, "remote sensing" has evolved to include a vast array of sophisticated instruments collecting information from both aircraft and satellites. One relatively recent advance in remote sensing is spectral imaging, which involves acquiring images at multiple electromagnetic wavelengths simultaneously. The wavelength information can be interpreted as spectral "signatures" of materials on the ground, allowing analysts to identify them remotely.

How spectral imaging works:
A spectral image can be thought of as many images combined, as shown here. For any location on the ground, the measurements of all the simultaneously acquired images are combined to form a "spectrum," whose bumps and wiggles are indicative of different materials.

Although HSI provides an "image" such as people are used to seeing, the spectral information is what makes it unique. The more spectral information available, the better defined the wiggles, and the more accurate the identification of materials.

Most commonly used today are multispectral images (MSI), which are available from aircraft and from satellites such as LANDSAT. Hyperspectral imaging (HSI), on the other hand, contains far more spectral information than does MSI, making it much more information-rich but also -- until recently -- much more difficult to process and analyze. HSI has the potential to transform numerous industries, such as agriculture, environmental monitoring and restoration, the oil and gas exploration, and military support. However, man-hours involved in processing and analyzing HSI data can be excessive, often requiring custom software and in many cases cost-prohibitive expert labor.

Now, as a result of a new SAIC data analysis engine, Abacus®, time to exploit hyperspectral image data has been dramatically reduced. Abacus makes information from HSI readily available, and can expedite critical decision making processes. Abacus automatically determines the subpixel proportions of materials in an HSI dataset and transforms the data into readily interpretable information layers that can be imported into a geographical information system (GIS) database. On a typical HSI data set, Abacus reduces the time and effort required from two to four days of an experienced analyst's time down to approximately 30 minutes of automated processing.

Applications of hyperspectral imaging data can support a diverse range of scientific and business operations. Among these, HSI data offers enormous potential for routinely creating geologic and soil maps, monitoring forestry and vegetation changes, and documenting hazards and the aftermath of natural disasters. Specifically, the high definition of HSI has recently received popular media attention because of its use to help winegrowers harvest their crops with more precision. By identifying plant health and maturity, vintners can establish optimum harvest schedules that ultimately yield high quality wine. The same methodology holds true for using HSI data to non-destructively detect bruised fruit in large orchards, and also aid law enforcement officials in detecting marijuana plants from airborne platforms.

For all of the applications outlined above, Abacus offers readily interpretable, automatic analysis of HSI. Because of its consistent and stable data interpretation, Abacus is a superior spectral image processing tool. This is ideal for both government and commercial applications where quality assurance is a key factor. To ensure its accuracy, an independent validation was conducted on Abacus by NASA Stennis Affiliated Research Center at Brown University.