Font Size: a A A

Research On The Segmentation Technology Of Remote Sensing Images Oriented To Feature Extraction

Posted on:2014-04-01Degree:MasterType:Thesis
Country:ChinaCandidate:J LiangFull Text:PDF
GTID:2310330536466597Subject:Photogrammetry and Remote Sensing
Abstract/Summary:PDF Full Text Request
The modern remote sensing technology has entered a new stage of dynamic,rapid,accurate,multiple means to provide various types of earth observation data,information of the target obtained from the remote sensing images has been widely applied in the national economic production and military intelligence field.The extraction of linear and planar features in the remote sensing images could provide a solid foundation for it.Image segmentation technology based on characteristics of each regions divided images into multiple regions,a better linear and planar feature extraction result can be got by judging these regions based on its shape character.For the significance of linear and planar features extraction,based on the realization of remote sensing image closed edge detection,the paper focused on the reaserch of segmentation technology of remote sensing images oriented to feature extraction.The major works and innovations of this paper are listed as follows:(1)Studied and summarized the situation and development trend of current remote sensing image segmentation technology,analysed the significance and research methods of linear and planar features extraction;(2)By analysis of edge detection technology and edge linking methods,proposed a method to detecte the closed edges of image,lay the foundation for reaserch of segmentation technology of remote sensing images oriented to feature extraction;(3)Sketched construction technology of the image pyramid,proposed the multi-scale segmentation technology based on image pyramid,and by the merger of the multi-scale segmentation results to optimize the result of segmentation;(4)Introducted the texture feature of several typical features in the remote sensing images;analied several kinds of statistical methods and texture similarity measure methods,through merging regions in the multi-scale segmentation results based on consistency texture to obtain the more optimal segmentation results.(5)Researched the shape feature of linear features in the remote sensing images to primary apply the segmentation results,realized the extraction of linear and planar features in the images.
Keywords/Search Tags:Feature Extraction, Image Segmentation, Closed Edge, Texture Feature, Shape Feature
PDF Full Text Request
Related items