Inland Waterway Target Extraction Based On Region-Line Primitive Association Framework

Inland waterway information management and detection is the important reference for ensuring the waterway transportation flow and the social traffic development,also is the necessary means for improving the capacity of inland navigation services.The traditional method of acquiring channel information is usually done by manual measurement,which is not only inefficient,but also Time-consuming.Therefore,how to quickly and efficiently obtain channel information is of great practical significance to the channel construction and water resource utilization.With the development of the remote sensing technology,high spatial resolution images can accurately refer to the real-time ground information of research areas and target areas.Object-based image analysis method is an advanced technology for high spatial resolution image processing and analysis.However,there are still some defects for the conventional object-based image analysis technology in the technical framework of segmentation and identify.For example,most of the methods do not make full use of the important classification clues such as the edge information.In this paper,a channel information extraction method based on region-line primitive association is proposed.Main work of this paper includes:(1)Extract region-line primitive and build model for the region-line primitive spatial correlation,by using the space relationship.Firstly,the method of image segmentation based on hard-boundary constraint and two-stage merging is used to get feature primitives.The phase-grouping method is applied to obtain line primitives.Then,the spatial correlation between region primitive and line primitives is obtained by using the region-line primitive association model,and the region-line primitives association features are extracted,which is prepared for the subsequent image analysis.(2)Object detection based on mathematical morphology.Firstly,the water area can be extracted by using the method of support vector machine(SVM)supervised classification on the basis of image segmentation,which is based on the analysis of the position and adjacent water features of typical artificial buildings(ports and bridges)in high resolution remote sensing image.Then,the suspected target area on the boundary of water can be detected by using mathematical morphology method.(3)Object extraction based on region-line primitive association rule constraint.In order to refine the morphological results of the extraction results,to achieve accurate extraction of ports and bridges,on the basis of the suspected area of the target,a set of rule constraint on the patch line density,the approximate vertical line,the straight line length-width ratio and the curvature of the axis is proposed,according to the positional relationship among the target,land and water,as well as the shape features of ports and bridges.Experimental results show that the algorithms in this paper can effectively and accurately implement the extraction of ports and bridges,while reflecting its advantages.