| Infrared imagery plays an important role in many fields,such as infrared search,recognition and tracking system,due to the advantages in working in all-weather conditions and hardly influence by climate.Because of the long distance,low energy and complex background consisted of random noise and non-stationary clutters,the moving small target only occupy very few pixels in the infrared images sequences,resulting in lack of detailed structure information or available textures.Therefore,moving small target detection is a challenging task,especially under complicated backgrounds.In this thesis,two algorithm based on temporal and spatial is studied to work out this problem.The main contents in this paper are as follows:In the chapter two,based on the feature analysis of background,clutter,noise and target,we model the background,clutter and target,which makes theory preparation for the propose of the algorithm.In the chapter three,we come up with a temporal variance top stagnation points(TVTSP)algorithm.Based on the temporal profile variance,filter connecting lines of the top stagnation points is designed to recognize the difference between targets,background and clutters,and then bring out the segmentation of targets.In the chapter four,a temporal and spatial fusion algorithm is proposed.The algorithm uses modified mean filtering based on the related information between adjacent pixels and the saliency of the targets,to restrain the background clutter.Afterwards,based on the target information between the frames,we adopt a modified temporal profile minimum filtering to make a further step on the suppression of the background clutter.The simulation experiment contrasted our algorithm with other classic algorithm,and the result shows that the temporal and spatial fusion algorithm is effective in improving signal-to-noise ratio.This paper has come up with two target detection algorithms,which support the advance in infrared target detection under complex background. |
PDF Full Text Request