 |
The work described in this paper deals with a non-parametric technique for target recognition. This method makes no assumptions about the underlying statistical processes for the data. Processing of the unknown targets begins with a structural analysis of the radar scattering centers making up the radar signature. The energy, location, and width of the scattering centers are parameterized. This symbolic description is then used to represent each target. The unidentified symbolic list is then used in the comparison process between known preprocessed templates and the unknwon object in question. The hypothesis evaluation focuses upon comparing similar scattering centers between the unknown object and the known templates using proximity constraints. A match confidence estimate is computed for each hypothesis based upon scattering center non-correspondences, energy differences, and matched location offsets. The original work was created using Mathematica and was designed for SAR imagery. The code has since been modified to handle UHRR data and has been converted into Khoros, the visual programming system. Keywords: Radar applications, signal processing; radar cross-sections; feature extraction; visual programming; symbol manipulation
|
|
For the newest resources, visit Wolfram Repositories and Archives »
