| The traffic force of Chinese people’s armed police armies have been officially bring into National emergency rescue system since 2013.Since they have been become emergency rescue national team,they play real role for the road repair in combat to the earthquake and carry out relief work for many times.But in the road repair procedure,due to the traffic and communication obstacle,we can not get the accurate disaster situation,so it has been influenced the rescue operation,if we can fast and precious obtain the road damage information,it will have a much significance in road rescue and emergency rescue.In recent years,with the rapid development of UAV remote sensing technology,it bring quickly obtained,bring the estimation of road damage information come true.UVA has better characteristic of flexible,simplicity,cheap,non-cycle limit,it can play a huge advantage in the complex environment of post-earthquake.The image-forming principle of optical remote sensing image is likely to the people’s eye,and using proper method analyzed the optical remote sensing image can give the emergency rescue officers a lot of road damage information without training.Based on this characteristic,this paper have interested research of UVA high resolution optical remote sensing image for the application of post-earthquake road repair.The research content hase been described as follow:Firstly,we research the post-earthquake UVA image acquisition methods and disaster area large scene image montage algorithm.Aim at the conventional UVA loop pattern can not satisfy the require of post-earthquake road image obtain,and make an improved pattern of multi-channel and multi-view loop road,this pattern has a better efficient and better adaptation for using in earthquake analysis.The aspect of image montage,aim at the conventional SIFT feature points image montage algorithm was ineffective for the high resolution optical images,we do improvement on SIFT features extraction and image rectification,the experiment results show that our proposed improved algorithm can effectively improve the precision of image montage,it provides the safeguard for the next step of road extraction.Secondly,in the tortuous road extraction algorithm of large scale high resolution optical images,there are a lot of algorithms,different algorithms are aimed at different specific scene for good effective,and many algorithms are used for straight-line and low curvature road.In this paper,we combine the requirement of timeliness of rescue,set out from the research of image cutting,first,we used the region road network information and large scale image interactive preliminary locate the road position,used Otsu image cutting algorithm extract the road region,then used image denoising algorithm,road reconstruction and refined to get the single-pixel road.Experiments show that this algorithm can extract tortuous road in a certain extent,but it also have some defect.Our paper also have more analysis for the reason of extracting defect,summarized the interference road type and influenced situation,and formulate some useful measures,it provided the data for road damage detection.Lastly,in the research field of road damage degree judgment,there are also lots of algorithms,the specific algorithm usually for specific data,in the first step,we compared some common damage detection algorithms,and chosen the simple difference method to detect the damage region,and then combined the road extraction result to analyze the adaptability of detecting result.Introducing and analyzing the represent patterns of damage type and image,and combined the difference algorithms’ result to complete damage judgment.Finally,by formulating and quantization the damage index,we complete the road damage degree classification and judgment. |
