Research On Intensity-based Image Matching Method

Image matching based navigation composes with other navigation methods has higher precision and reliability. This paper is dedicated to the algorithm research of image matching applied to navigation. After analyzing and comparing several traditional intensity-based matching algorithms, a new matching method combing the improved mean-residual normalized intensity correlation and wavelet multi-resolution layer matching is presented and proved by experiments. The research contents of the paper include:1 )This paper introduces the basic theories of image matching, including concept, mathematics description, process flow, similitude measurement, research contents, existing matching algorithms and factors that affect matching performance.2)This paper discusses matching pre-processing. The error factors such as intensity distortion, noise and geometry distortion will affect matching performance, so the paper gives the opposite processing methods. Information such as attitude angle of aircraft is used to emendate the large-geometry-distorted image, so the research of this paper is based on the hypothesis that images used have small geometry distortion and invariant moment feature is used to reduce this distortion.3)This paper compares several intensity-based matching algorithms, through experiments analyzes their respective merits and demerits, and improves the computing method of NNPROD as result to reduce computing quantity.4)This paper expounds the basic theory of wavelet transform , makes use of the characteristic of wavelet multi-resolution to delaminate the image and explain in detail the existing problems in matching. Experiment results show that under the noise, small geometry distortion and intensity distortion, the matching method given by this paper which combing the improved mean-residual normalized intensity correlation and wavelet multi-resolution layering has good matching performance.