Research On Infrared Small Target Detection Algorithm Under Different Cloud Back Ground

Due to the small number of target pixels,no size information and the influence of the complex battlefield environment,the infrared small target detection problem has been a hot spot and difficult point in the field of infrared image research.Based on the existing algorithms based on mainly for different infrared small targets detection in cloud background technical studies,research contents are as follows:(1)Study on single frame image detection algorithm for infrared small targets.To solve the problem of interference caused by complex sky background clouds,in the analysis of four kinds of characteristics of infrared image under the sky clouds on the basis of a small infrared target detection algorithm based on local difference,through the establishment of mathematical model,the difference between pixels in infrared images based on local quantitative analysis,the difference between pixels.The establishment of target detection threshold.To validation local poor value algorithm of superiority and adaptability,respectively on four species sky clouds Xia of small target do detection simulation,and and traditional algorithm of detection results do compared,simulation proved,local poor value algorithm can accurate to detection out different clouds in the of small target,compared and traditional of detection algorithm,the algorithm has virtual police less,anti-clouds interference sex strong,adaptability good,advantages.(2)Research on algorithm of small target detection in image sequences.Detection of small target in image sequences,is presented in this paper are based on local difference and pipeline filter combination detection algorithm,do a single image detection using local difference algorithm,then pipe filter algorithm for single frame image results for secondary judgment,concluded that the real target.In the decision process,introduced a new sentencing guidelines--continuous counting standards.It effectively solves the shortcomings of the traditional pipeline filtering algorithm,which is poor real-time performance and lack of strain.The two sets of images are simulations,results show that the algorithm can accurately detect small targets in image sequences and sudden target in image sequences have a strong ability to,has a high practical value.