遥感影像亚像元定位算法分析-analysis of sub-pixel location algorithm for remote sensing images.docxVIP

AbstractDue to the limitations of the spatial resolution of the Remote Sensing Image(RSI), mixed pixels are ubiquitous in RSI. This phenomenon affects seriously the level of application, and become an important obstacle for the quantitative analysis of RSI. Sub-pixel Mapping technology can get more detailed spatial structure within the mixed pixel, provide higher spatial resolution information, and reflect more accurately details in RSI. It is helpful to improve the accuracy of RSI target recognition ans terrain classification.In this paper, we, firstly, analyse researches in sub-pixel mapping field, and then focus on two categories according to different application purposes. The first one is sub-pixel mapping method based on single target, who’s purpose is to restore the shape and contour of interested targets in a higher resolution data, by using the abundance of low spatial resolution image; The another is multiple features mapping, who’s result is a image classification with a higher resolution through processing low spatial resolution image with the sub-pixel mapping algorithm.For the first application, we propose a algorithm, named Based on the boundary guidance of the single target mapping methods, by taking into account the characteristics of low resolution targets, sub-pixel spatial dependence and the geometric features. Experiments show that the algorithm can effectively improve the ability of single target mapping, and provide a more accurate target geometry information.And for the second application, another method is proposed, named Sub-pixel Mapping Algorithm Based on Weighted Gravitation Model, based on the gravitation model. Experiments prove that this method can effectively increase the accuracy of multiple features mapping, and improve the classification accuracy of surface features.This paper also analyses and compares evaluation methods for sub-pixel mapping results, and then gives the way of that how to choose a mapping accuracy evaluation f