Object Stereo Information Recovery Based On Multisource Remote Sensing Information

With the quickly development of remote sensing technology and the improvement of various applications requirement, spatial resolution and spectral resolution of remote sensing earth observation data has been higher and higher, which makes multisource remote sensing data contains more and more abundant and detailed information. At the same time, compared with traditional areas-based analysis, object-based analysis received more and more attention from scholars. And people are no longer satisfied with two-dimensional information provided by the remote sensing image. Therefore, how to make use of multisource remote sensing information to build highly reality stereo model has become a pressing problem in remote sensing field.This paper mainly aims at the object of multisource data of high spatial resolution satellite, aerial remote sensing images and Li DAR data, using texture information in remote sensing images and stereo structure information in Li DAR data to realize the visual object stereo information recovery.First of all, in order to obtain each surface texture of object in satellite and aerial images, based on a new texture segmentation method called ORTSEG(Occlusion of Random Texture Segmentation) algorithm, we proposed an improved ORTSEG algorithm. We verify our algorithm through comparative experiments with other three algorithms as well as quantitative experiment. The segmentation results show the superiority of this improved ORTSEG algorithm, especially the segmentation results of object which has almost same texture surfaces and obvious surfaces edge. This lays the foundation of the relevance of object surface texture and stereo structure information in next part.Secondly, using the idea of image registration, we propose an object surfaces texture slope geometry correction method. In order to obtain front view texture of object we correct the surfaces texture from the texture segmentation. Then based on the top stereo structure point cloud data of satellite and aviation object from Li DAR data processing, we propose a relevance algorithm of object surfaces texture and stereo structure information based on object top contour guidance. The experiment results show the excellent relevance result. And we verify our algorithm’s feasibility and universality. This paves the research of object stereo information recovery in next part.Finally, based on the two parts before, we propose an algorithm of object stereo information recovery based on relevance of surface texture and stereo structure information. In order to make object stereo structure into flat plane and recover object stereo structure information, we triangular meshing the whole stereo structure of satellite and aviation object. Then we recover object surface texture information using the corresponding relationship of surface texture and stereo structure built in relevance part. Then we prove the fine and universal of this method through qualitative and quantitative experiments and realize the visual object stereo information recovery.