Multiple location SAR/ISAR image fusion for enhanced characterization of targets

The main focus of this paper is the development of fusion strategies for multiple location synthetic aperture radar (SAR), and inverse synthetic aperture radar (ISAR) images. The techniques being developed are to be used in conjunction with super-resolution and target identification strategies for non-cooperative target recognition (NCTR). Multiple location processing has the ability to provide improved image quality as well as target detection and classification capabilities since the different aspects or "looks" can provide additional clues about the shape, dimensions, and special features of a target. Many traditional SAR/ISAR processing techniques seek to maximize the instantaneous SNR for a signal in the presence of additive noise. Unfortunately, these techniques do not directly address the recreation of an image with minimum mean squared error between the reconstructed SAR image and the reflectivity map of the actual scene. This paper examines techniques capable of improving the probability of object detection within an image generated via spatial fusion. The strategies focus on image level fusion of the SAR/ISAR data. Canonical SAR/ISAR data is used to validate and compare the fusion results. Preliminary results using DARPA's MSTAR database are also presented.