SAR Image Registration And Target Recognition Via Correlation Matching

Image of Synthetic aperture radar(SAR) systems are all-weather, all-time, high resolution and strongly penetrating power, thus image processing and understanding is a subject of a area which is profound and meaningful. In the recent decades of years, Domestic and foreign scholars enrich the theoretical of this discipline and accuracy may outstanding academic achievement with continuous efforts, innovates and practices.Registration and target detection and recognition in area of image processing of SAR image has very important position. Registration which is accurate and efficiet become a key step of image processing. The accuracy of target detection and recognition has also become an important part of the SAR image understanding.This thesis focuses on the image registration and target detection and recognition algorithms that the synthetic aperture radar image is impacted with the speckle noise.(1) In this thesis, firstly proposing a SAR image registration method using correlation function which is based ratio term. This work studied SAR image speckle noise characteristic and found that the traditional form of function based product term can hardly suppress the influence of the speckle SAR noise in a registration strategy, thus, we pick up a new correlation function based ration term which is more suitable for the process of registration. Then we use a simple but accurate priori mean to do a preprocessing in addition. Experimental results show that the new correlation function based ratio is more suitable for SAR image processing. The adding of priori is able to overcome the impacts of SAR speckle noise.(2) Designing an airport aircraft target detection and recognition algorithm which is based on template matching. Center of this method is excluding the object which is interference process of algorithm. The method can remove the points, lines and surface targets that are false alarms. Combing with the template library, the goal that apron and aircraft detection and recognition be completed. This algorithm achieved a higher detection rate over 90%, which can meet project requirements.