Rocket plume image sequence enhancement using 3D operators

Reported here are the results of tests of five 3D image operators in eleven configurations to determine which is best for the enhancement of image sequences of rocket plumes. The operators enhance images both for visual inspection and for radiometric characterization. The tests used real and synthetic imagery to evaluate the operators under a variety of conditions. The results suggest the best choice of operator depends on the grayscale distribution of the noise in the image sequence. In terms of signal-to-noise ratio (SNR), long duration median filters with moderate spatial support appear to work best overall. The results show that successful enhancement for visual inspection requires the original image data to be remapped prior to filtering. This need not destroy the radiometric information in the imagery. Some attributes of rocket plume image time sequences and the noise in them are described. The remapping procedure is detailed and the operators tested are described. The methodology is explained and the results of the numerous tests on synthetic imagery are presented. The application of the procedures to two real rocket plume image sequences is described.