The Study, Based On Image Matching Navigation

Image registration is to find correlation from different images of the same view. The method of image match in navigation system is analyzed. High precision navigation technique is important in military and civilization. Image aided navigation is used in many navigation systems to improve the precision of the system. Image match is critical technique in image aided navigation. Image match system finds correspondence between real image and base image and then outputs the position parameters to the aircraft. The edge extraction, computation of statistical feature, image match techniques and design of image aided navigation simulation system are covered in this thesis.The edge extraction methods are researched firstly. Some common edge extraction methods are discussed and analyzed. After that the edge extraction methods of color image is discussed since the images captured by some sensors are color images instead of gray images and the difference of edge extraction in different color space.The moment invariant feature is chosen as invariable feature of the image also. The moment feature is a kind of statistical feature. It presents the feature of the image in a whole. The geometry moment and the zernike moment are discussed here. Zernike moment is invariable when image rotates and it is orthogonal, so it is suitable as invariant feature of the image. The structure moment is used for edge image. The computation of structure moment based on zernike moment is presented here.Image match methods consist of two kinds of methods, area-based and feature-based. Feature-based is often used in image aide navigation system because the sensors and the time of image captured are different. The edge based technique is mainly researched.The match of two images needs to be done after got the features. The elements of image match are similarity measurement, search strategy and search space etc. The discussion is restricted to affine transformation here. The moment invariant and hausdorff distance is researched as similarity measurement. The GA strategy, the pyramid search strategy and wavelet based multi resolution strategy are analyzed as search strategy. Finally the wavelet based multi resolution strategy is chosen after experiment and comparison.Finally a simulation program is designed and implemented as a simulation tool ofimage navigation system. The software can simulate the precision and functionality of image navigation system. It consists of several modules which are route selection module, real image creation module, image match module and result presentation module. The functionality and performance of the feature extraction method and the image match method can be reviewed directly and quickly via simulating the work flow with the simulation software.