Road Detection For High-Resolution SAR Image With The Sketch Map And Priori Constraints

Synthetic Aperture Radar(SAR)is an active radar.The most significant advantage of SAR imaging technology is that it is not affected by the weather or the environment,highresolution remote sensing data can be acquired all day and all weather.Because SAR technology has very obvious advantages,therefore,it has been applied to many fields such as military,marine,and civilian.As an important land feature,roads play an important role in transportation and connectivity.Therefore,how to more accurately extract road networks from SAR images is gaining increasing attention,it has become a hot topic in recent years.Due to the complexity of the background environment and the diversity of road types,as a result,road detection in SAR images has problems such as a large false alarm area and inaccurate target positioning.After analyzing and studying the problems faced by current SAR images in road detection,a road detection method for high-resolution SAR images based on sketch maps and priori constraints is proposed.The innovations are as follows:(1)A sketch segment extraction method based on spatial scalability and grayscale distribution connectivity that can represent roads is proposed.In the sketch map of the acquired SAR image,for sketch segments that can represent a variety of semantic information,first of all,we should extract the sketch segment that can represent the road.In this paper,first of all,using the priori Constraints and statistical information of the road,the preliminary screening of the suspected road sketch segment is performed;Then the information in the semantic space and the pixel space of the SAR image is exchanged,and the rules for calculating the spatial extent of the sketch segment and the degree of connection of the gray scale distribution are designed.By combining the two features,a set of sketch segments representing the road can be obtained.After a simulation experiment,this method can screen out a large number of false alarm areas,achieve validation of the validity of sketched segments that represent roads.(2)A sketch segment heuristic connection method based on MRF chain is proposed.Due to noise,blocking and other interference factors,roads in SAR images map to sketch maps are expressed as a sketch segment broken.For this issue,We use the unique ductility of roads to use the extraction algorithm that can represent the sketched segment of the road,that combine the two features of spatial extension and gray distribution connection degree to obtain the connection condition probability between sketched segments,based on the idea of Markov chain,calculate the maximum connection probability to determine the current sketch segment connection status,and the next sketch segment that is suitable for connection.Through experimental simulation,this method complements the broken part of the SAR image,achieve the exact connection of the sketch line segments that fit the road.(3)A method of locating a road area based on super-pixel segmentation is proposed.As the road is a band structure with a certain length and width,therefore,it is necessary to position the lateral area of the road.For drawing lines that have been connected to represent the road,through watershed super-pixel segmentation,using information about the area and position of the super-pixel around the sketch line,determine the relative position of the road area and sketch line,and use Hough transform design algorithm to determine the road edge.Through experimental simulation,this algorithm can accurately locate the road target,the final road area is better determined based on the relative position of the road edges.