Dr. Michael R. Hawks, Ctr.

• Bio Home
• Publications
• Work History
  Professional Memberships
• Research
  Research Interests Expertise
• Teaching Interests
  MS Thesis Advised

Research Interests

• HSI Sensor Development: hyperspectral imaging (HSI) is a proven method for material classification and identification. The product is a three-dimensional data structure of spectral information at each pixel in a 2-D image. In order to record this 3D data structure on a 2D sensor array, conventional HSI instruments require many data frames, which means it can be very difficult to use them to measure any scene that changes over the timescale of the measurement.  AFIT is developing two different high-speed HIS sensors.  The first, based on chromotomography compresses the number of frames, and therefore the time needed to capture the data.  The second is a true snapshot hyperspectral imager.
• Detection & Tracking: surveillance sensors continue to grow in both number and in array size. This has resulted in a huge increase in data volumes, without a corresponding increase in analysts. Automated methods to detect moving, unresolved targets can augment the analyst workforce, but current methods are either too slow or insufficiently sensitive.  AFIT is researching improved ways to automatically extract more information from existing data sources.  The two main efforts underway now are (1) exploring better detection algorithms to reduce detection thresholds and (2) fusing data from multiple sensors to suppress noise-driven false alarms.
• Disturbed Earth: a reliable field-deployable system that can detect shallow-buried objects like mines is still a high priority problem. Some studies show small but detectable traces observable with thermal imaging and polarimetric imaging, but  neither detection method  is sufficiently sensitive and reliable on its own. AFIT is conducting experiments using our unique Polarimetric Hyperspectral Imaging Camera System (PHySICS) to explore possible benefits to combining multiple sensing methods.
• EO/IR Signature Modeling: Development of new detection and tracking algorithms requires testing against data sets with known truth data, which can be very difficult to get for some sensors.  AFIT has developed physics-based scene simulation tools to produce realistic simulated data with full control over target, background, atmosphere, and sensor properties. To make the data more realistic, we are working to develop fast signature modeling tools to estimate the reflected and emitted signatures of aircraft or other targets.
• Atmospheric Transport / Atmospheric Correction: the signals observed by hyperspectral remote sensing instruments contain information about the target, but also about the atmosphere and other illumination sources (e.g. sunlight).  In order to improve target detection/identification, we are exploring methods to improve capability to quickly correct for atmospheric transmission.