Developing Algorithms For Small target Detection Using Infrared Images

Abstract

This thesis work has extensively explored the various methods for detecting a dim and small target in a noisy background. Many applications such as military system, remote sensing, infrared search and tracking and medical imaging. newlineSo much of work has been done in this direction in the past, but still it is an open issue that the better algorithms need to be developed. The problem statement of this thesis work focus on developing algorithms for detect before track approaches. Moreover, the algorithm developed should be robust computationally efficient. newlineThe existing infrared patch image IPI based methods are efficient than the other baseline methods, but still these approaches suffer from the problem like l1norm minimization which makes the estimation of background image inaccurate. Global constant weighting issue also need to be addressed. So, methods need to be devised to reduce the computation Burdon of these algorithms. newlineIn this research work, we have achieved the following objectives newline Detailed explanation and comparative analysis of existing state of the art methods has been done. newlineA, method Reweighted infrared patch image model for small target detection based on non convex lp norm minimization and TV regularization has been proposed.An approach for Infrared small target detection based on Non convex Triple Tensor Factorization has been proposed.An approach for Infrared small target detection based on Bi Nuclear norm minimization has been proposed.A method for Infrared Small target detection based on Phase Fourier Spectrum Transform have been proposed.A method for Infrared Small target detection based on Fractional directional derivative and Phase Fourier Spectrum Transform has been proposed.At the end, a method for Infrared Small target detection based on non convex convex lp norm minimization has been proposed newline newline