Apply online to license this technology
A scientist at the Army Research Laboratory has invented methods to improve image reconstruction. The patented technology is available via patent license agreement to companies that would make, use, or sell it commercially.
3D terrain mapping technology has been used to provide terrain visualization, with recent improvements being made in resolution, accuracy, quality, and amount of area covered. Nonetheless, the transformation of 3D maps from low resolutions, appropriate only to large areas, to high resolutions, useful at much smaller scales, requires a new approach to terrain visualization. The shortcomings of current terrain visualization offering include those experienced in visualizing data acquired through sensors, such as LIDAR devices, that have been employed for applications such as detecting and tracking people and vehicles.
Imaging techniques have also incorporated or adapted methods for detection of surface variations. When imaging smaller-scale scenes, conventional devices tend to under-sample small-scale features, such as foliage, fences, railings, and light poles. These features are difficult to identify and distinguish using commonly employed terrain visualization techniques. Yet, it is important for a visualization method to accommodate this under-sampling and to differentiate between these under-sampled objects and larger, smoother objects like vehicles.
To address these problems, ARL has developed a texture-based segmentation algorithm designed to eliminate the small gaps in surfaces that are basically smooth – without sharp changes in elevation. This algorithm identifies locally relatively smooth areas and warps the squares for these areas so that the edges meet and the surface is continuous. Areas that are truly rough, such as trees, are left as disjoint squares.
One important feature of this procedure is that there is only a small difference between the squares (or pixels) before and after the algorithm is applied, so algorithm errors are not conspicuous. Thus, the algorithm does not need to be extremely reliable and can be fast and simple. The algorithm may occasionally leave a gap in a smooth area or occasionally join several squares in a tree; but such errors are almost imperceptible. The visual impression is to eliminate most of the small gaps in smooth surfaces making more striking the contrast between roads, grass, and buildings that are smooth and disjoint under-sampled objects like trees.
- Visualization technique portrays both small and large features and as well as natural and manmade features
- Technique is applicable for both horizontal and vertical plane imaging
- Businesses can commercialize the technology by licensing U.S. Patent 8,249,346 from the Army
- TechLink guides businesses through evaluation and licensing; services provided at no cost