The Theory And Technology Research On Infrared Target Detection And Recognition

In this dissertation, some researches are carried out for infrared target detection, infrared target recognition and the algorithm of infrared image processing. The application of space-time filter is paid more attention to detecting infrared target and moving detection analysis for infrared target. And then a study of infrared target was made based on gray scale estimating and research on stepwise match for recognition of infrared target. At last, the algorithm is analysed for recognition of infrared target based on mark and neural network.In order to detect infrared target and restrain the background, the space-time filter from time domain to spatial domain is adopted. An adaptive spatial filter is applied to decrease image noise. According to the target edge caused by gradient and the possibility of losing useful information after the difference of time domain filter, detecting the target edge is proposed before time domain filter and perform the interpolation compensate by the information of neighborhood pixel after time domain filter. The noise of time domain was not related to that of spatial domain mutually. The synthesis combining spatial domain filter and time domain filter is the image of high S/N ratio. It is effective for space-time filter to detect the infrared target, also is suitable for complicated background.Optical flow algorithm, gradient algorithm and grey system estimating are applied to solve the moving detection problem of infrared target. The simulation and the experimental results show that flow algorithm and gradient algorithm which being in combination with each other have engineering practicability. That is to say, optical flow algorithm finds the motion rule of the target (under the condition of forming optical flow), and then gradient algorithm measures that accurately. Grey system estimating is adopted to predict the interframe motion trajectory of moving target. These algorithms could have certain constructive significance to detection of moving target. Meanwhile, the applicability of detecting moving target has been inspected as far as possible. Aimed at detection of infrared target based on gray scale estimating, an algorithm for estimation of improved maximum entropy was proposed. The flexible factor and mandatory factor of the entropy were proposed for the first time. The mark was assigned to the corresponding target, which gives evidence to target detecting. Gray scale estimating and filter were carried on based on pixel morphology operator.According to meeting the requirement of maximum a posteriori probalility of image segmentation based on Markov random field, the simulated annealing algorithm could converge at local optimal solution, which was based on game theory. After simulated annealing, stepwise match of the moment-fourier descriptors feature matrix was carried on by characteristic of moment function. As a result, stepwise match was in combination with simulated annealing segmentation pretreatment and it has a good recognition rate for infrared target as far as possible.At last, the theory of recognition of infrared target based on mark and neural network is studied. By selecting an actual infrared target, the recursive mark is assigned to the target based on made of segmentation pretreatment on the image. According to the recursive marking as input sample and the centroid of the target for output, the BP neural network is constructed. After the anti interference training to the network, it carries on a correlation matching of alterable resolution by the corresponding centroid characteristics of the infrared target. Not only does it improves the recognition rate but also enhances the real-time imagery processing enormously.The research achievement of this dissertation is that simulation results go deep into the image processing of infrared target, which is a part of the system of detection and recognition of infrared target. The effective algorithm for detection and recognition of infrared target under complex circumstance could have directory significance to engineering practice.