Sparse Array RF Imaging for Surveillance Applications

Abstract

Techniques are provided for sparse array RF imaging for surveillance applications. Objects in a three dimensional (3-D) image-data-set obtained from multi-static radio frequency detection data are classified, for example, as human or non-human. One or more geometric image features are extracted from the image-data-set that support a target classification process; and the one or more objects are classified as a threat based on a parametric evaluation of the extracted geometric image features.

We claim: 
     
         1 . A method for classifying one or more one or more objects in a three dimensional (3-D) image-data-set obtained from multi-static radio frequency detection data, said method comprising:
 extracting one or more geometric image features from the image-data-set that support a target classification process; and   classifying said one or more objects as a threat based on a parametric evaluation of said extracted geometric image features.   
     
     
         2 . The method of  claim 1 , wherein said extracted geometric image features comprise one or more of a height of an image center of mass above ground, an image occupation extent, vertical and horizontal aspect ratios, image alignment with target direction and orientation of the image with respect to a track heading vector. 
     
     
         3 . The method of  claim 1 , wherein said extracted geometric image features comprise one or more of image surface area, discernible image appendages and position and target shape evolution through a sequence of images generated sequentially over a subset of a target track. 
     
     
         4 . The method of  claim 1 , wherein said extracting step further comprise the step of extracting one or more features from the image-data-set that characterize a scattering strength of a moving object at a selected image track point. 
     
     
         5 . The method of  claim 1 , wherein said classifying step implements one or more of neural networks, hidden Markov models, and multi-variable Gaussian classifier manifolds. 
     
     
         6 . The method of  claim 1 , wherein said classifying step is based on a classification score. 
     
     
         7 . The method of  claim 6 , wherein said classification score indicates a confidence of the classification. 
     
     
         8 . A system for classifying one or more one or more objects in a three dimensional (3-D) image-data-set obtained from multi-static radio frequency detection data, said system comprising:
 a memory; and   at least one processor, coupled to the memory, configured to:   extract one or more geometric image features from the image-data-set that support a target classification process; and   classify said one or more objects as a threat based on a parametric evaluation of said extracted geometric image features.   
     
     
         9 . The system of  claim 8 , wherein said extracted geometric image features comprise one or more of a height of an image center of mass above ground, an image occupation extent, vertical and horizontal aspect ratios, image alignment with target direction and orientation of the image with respect to a track heading vector. 
     
     
         10 . The system of  claim 8 , wherein said extracted geometric image features comprise one or more of image surface area, discernible image appendages and position and target shape evolution through a sequence of images generated sequentially over a subset of a target track. 
     
     
         11 . The system of  claim 8 , wherein said one or more geometric image features are extracted from said image-data-set by extracting one or more features from the image-data-set that characterize a scattering strength of a moving object at a selected image track point. 
     
     
         12 . The system of  claim 8 , wherein said one or more objects are classified as a threat using one or more of neural networks, hidden Markov models, and multi-variable Gaussian classifier manifolds. 
     
     
         13 . The system of  claim 8 , wherein said one or more objects are classified as a threat based on a classification score. 
     
     
         14 . The system of  claim 13 , wherein said classification score indicates a confidence of the classification. 
     
     
         15 . An article of manufacture for classifying one or more one or more objects in a three dimensional (3-D) image-data-set obtained from multi-static radio frequency detection data, said article of manufacture comprising a non-transitory machine readable recordable medium containing one or more programs which when executed implement the following steps:
 extracting one or more geometric image features from the image-data-set that support a target classification process; and   classifying said one or more objects as a threat based on a parametric evaluation of said extracted geometric image features.   
     
     
         16 . The article of manufacture of  claim 15 , wherein said extracted geometric image features comprise one or more of a height of an image center of mass above ground, an image occupation extent, vertical and horizontal aspect ratios, image alignment with target direction and orientation of the image with respect to a track heading vector. 
     
     
         17 . The article of manufacture of  claim 15 , wherein said extracted geometric image features comprise one or more of image surface area, discernible image appendages and position and target shape evolution through a sequence of images generated sequentially over a subset of a target track. 
     
     
         18 . The article of manufacture of  claim 15 , wherein said extracting step further comprise the step of extracting one or more features from the image-data-set that characterize a scattering strength of a moving object at a selected image track point. 
     
     
         19 . The article of manufacture of  claim 15 , wherein said classifying step implements one or more of neural networks, hidden Markov models, and multi-variable Gaussian classifier manifolds. 
     
     
         20 . The article of manufacture of  claim 15 , wherein said classifying step is based on a classification score.