Spatial Filtering for Enhancing Point Targets in Images from a Space-Based Mosaic IR Detector,

Abstract : We examine by simulation the use of spatial filtering for enhancing the visibility of point targets and reducing the background clutter in imagery obtained with a space-based staring mosaic IR detector array. An experimental digital image-processing system was employed to interactively process Landsat II images containing simulated point targets. First, we obtained improved target detection and clutter rejection by using filters that produced considerably more high-frequency enhancement than the conventional Wiener filter. Second, convolution filters with large-area point-spread functions (e.g., those with 11 by 11 elements or those implemented with a 2-dimensional FFT algorithm) were more effective at attenuating the edges of extended regions of clutter (e.g., those corresponding to clouds, bodies of water, etc.) than were those with small areas (e.g., 3 by 3 or 5 by 5 elements). (Author)