For the segmentation of forward-looking infrared (FLIR) images, a new metho
d was proposed, which uses median subtraction filter to suppress background
and enhance targets, then uses the model-based segmentation (MBS) of FLIR
images to perform segmentation. A new computational method of compatibility
vector and initial probability was proposed to improve MBS algorithm, The
segmentation results with real FLIR images prove that the presented method
can give more precise and accurate segmentation of targets than MBS algorit
hm under low contrast and high noise conditions, and reduce the influence o
f background greatly.