Research On The Key Technology Of Imaging Navigation

Imaging navigation is a navigation method which associates two-dimensional map with real images.It is interposed between two and three dimensions,which is a complement to the two dimensional map navigation and a simplification of the three dimensional navigation.That has some advantages such as convenient data acquisition,low cost,rich geographic information and good performance in visualization.It is becoming a hot spot in the field of navigation.And it is of vital significance to study the key technology of imaging navigation.The paper carries out researches about the key technology of imaging navigation both from theoretical and practical aspects.The main work and contributions are as follows:(1)The image stitching technology based on features is studied to achieve a mosaic image according to the general process.Several commonly used feature extraction algorithms are analyzed and advantages and disadvantages of every algorithm by comparative experiments are obtained.Feature points matching methods of Normalization Autocorrelation and Euclidean distance are studied.Based on the classical RANSAC algorithm,an algorithm of purifying matching by topological constraint of feature points to speed up the RANSAC algorithm is presented.Experiments show that proposed method is more efficient,faster and perform well to the rotation and scaling of images.And image registration results show that the effect of image stitching is better through the homography matrix calculated by the topological constraint of features.(2)Technology on associating images with the location is studied.Related technologies of the unified of time reference are introduced,and the principle of image data associated with location is analyzed.A method to associate images with location based on time reference is proposed.The program of associating images with location is developed on the Visual Studio platform.(3)The technology of perspective transformation algorithms in imaging navigation is studied.The state-of-art view morphing methods and principles are analyzed and the mathematical model of view projection is deduced in detail.Experimental results show that because of multiple resampling in process of pre and post-transformation,the intermediate view is blurred and therefore it is not applicable for imaging navigation.To meet the needs of perspective in imaging navigation,a perspective transformation algorithm based on location and scene plane is presented,which utilizes the principle of homography induced by scene in computer vision.The formula of homography between different locations is deduced.It can effectively determine the view transform and avoid ambiguity of traditional view morphing algorithms.