Progressive Image Discrimination


The research team, led by Dr. Ken McGwire of the Desert Research Institute, has developed a highly interactive and informative mapping methodology that will allow you to continuously refine desired results from images used in Geomatics. Geomatics involves collecting, analyzing, and mapping geographic information that includes, but no limited to, the planning, mapping, and assessing of geographical areas for land use, transportation, forestry, mining, agricultural production, and invasive plant and water resource management.

Technology Summary

This technology is an interactive human/machine method of gradually refining the selection of areas of an image based on the desirable data content. This progressive discrimination method has been used with aerial photos taken that recognizes spectral signatures such as, but not limited to, plants, which would identify and/or separate invasive species. This technologies interactive tool in effect, would progressively isolate the desired target from similar spectra in an image. The highly informative mapping has been tested with aerial photos in the field of geometrics by collecting, analyzing, and mapping geographic information. This new progressive discrimination method simplifies the process of mapping specific features seen in satellite and airborne imagery while consistently matching or exceeding the current methods of experts. With this technology, non-expert users can easily negotiate very complex datasets like hyper-spectral imagery. When given only a few examples of the desired image features, progressive discrimination can map user-selected characters. With a very simple collaboration, the end-user can turn image data into accurate map information without the need for expert support saving money and time.

Features and Benefits

  • Simple data analysis technique allows the novice user to manipulate data creating a desired map with local knowledge and reducing 3rd party expert costs.

  • Progressive discrimination finds specific features that match desired examples yielding consistently higher classification accuracy than any current product tested with an expert analyst.

  • Eliminates current manually intensive methods that require extensive technical knowledge which translates to less errors, fewer time delays and money saved without having to purchase expensive specialized image processing software.

Potential Applications

  • Image manipulation

  • Land Use Planning and Transportation Development

  • Forestry Management – Invasive plant management, Inventory of current species of plants, logging management.

  • Mining – Determining the location of certain geologically rich areas, could drastically reduce efforts for locating additional minerals.

  • Water Resource Management – Detection and monitoring of current water supply.

  • Medical image analysis – Tumor identification

  • Emergency Response Allocation – Determining areas affected by natural disasters including floods, tornado, earthquake, etc. Used to dispatch emergency personnel to proper areas.

IP Status

DRI ID#: DRI07-001
Data analysis process
U.S. Patent Appln. No.: 12/532,582

Patent Information:
For Information, Contact:
April Sferrazzo
Account Tech
University of Nevada, Reno and Desert Research Institute
Kenneth McGwire