Extraction Of Damage Information In High-resolution Road Remote Sensing Image

The devastation of the earthquake will not only cause serious damage to the socio-economic and the environment,but also a great threat to life safety.The first time after the earthquake disaster situation will help rescue the command department to develop rescue work program,which will minimize the loss and threat.But the earthquake caused by road collapse,rupture and other damage,and its secondary disasters will cause serious road hindered,buried,etc.,leading to rescue workers,rescue vehicles can not enter the disaster area,rescue work will not be able to start.In recent years,with the remote sensing image of the spatial resolution of the increase and the sensor constantly updated,making the human exploration of geographic information easier.Through the remote sensing technology can be timely to the traffic repair department to provide road damage,damage distribution and other information,which is to reduce the impact of disaster and rescue casualties has a very important significance.Remote sensing system in the access to information by time,spectrum,space and resolution and other conditions,it is difficult to accurately observe and record complex and rich geographic information,and access to observation data will be subject to the atmosphere,clouds and regional complexity Degree and other factors will inevitably exist a certain degree of error.In this paper,the quadratic polynomial is used to smooth the contours of the distorted source in the image and to smooth the anisotropic diffusion model.The high quality remote sensing image is obtained and the reliable basic data is extracted for the follow-up information.For the original data in the acquisition process there is a certain error,this paper uses the quadratic polynomial tothe geometric correction of the distortion source in the image;then use the coherent enhanced anisotropic diffusion model for smoothing to obtain high-quality remote sensing images for subsequent information extract reliable base data.According to the different extraction of remote sensing information,the scale parameters of the segmentation are not the same.When the split scale selection is unreasonable,it will cause problems such as "under-division","over-division","edge-mismatch".Firstly,the fractal network evolution method is used to segment the original image.Then,the optimal initial clustering center is determined from the small-scale object of the pre-segmentation by using the global search ability of the particle swarm algorithm.In the small-scale object clustering The paper constructs the objective function with the information of the object space information and the object.The paper divides the given algorithm at different scale,and compares it with eCognition Developer 8.7 software and watershed algorithm.The experimental results show that the proposed algorithm has better segmentation results and can be used to segment the results of different scales,which reduces the over-reliance on the scale parameters by the multi-scale segmentation method.In the fuzzy classification of the segmentation object,the paper analyzes the damage road feature information,and introduces the weight coefficient according to the different characteristics in the classification,thus improving the weight of the main feature,and reducing the weight of the secondary feature.The experimental results show that the classification accuracy is higher when the different weights are given different weights than the ones that use the same weights.