Research On Road Extraction Based On High Resolution Remote Sensing Image

With the rapid development of remote sensing technology,the spatial resolution and temporal resolution of satellite imagery also have a huge increase.Meanwhile,High-spatial-resolution images are becoming increasingly popular for commercial applications.The remote sensing image technology has broad application prospects in intelligent traffic.Compared with traditional traffic information collection methods,vehicle information extraction using high-resolution remote sensing image has the advantages of high resolution and wide coverage.This has great guiding significance to urban planning,transportation management,travel route choice and so on.In this paper,the fusion image of panchromatic and multispectral from high-resolution remote sensing image is used as the data source,and the method of road extraction based on high-resolution remote sensing image is mainly studied.The specific research content and innovation work are as follows:(1)Research on road extraction algorithm based on spectral characteristics of land-cover.Firstly,this paper preprocessed the high-resolution multi-spectral and panchromatic remote sensing images acquired,and image fusion was taken.After that,histogram equalization and linear enhancement are performed for the fusion images,and the optimal threshold segmentation is performed.Meanwhile,according to the spectral characteristics of vegetation,water and other background,the vegetation and water information are extracted by normalized water index(NDWI)and normalized vegetation index(NDVI).Then,combining the optimal threshold segmentation result and the result of vegetation and water extraction,the road target after background suppression is obtained.Finally,the shape features such as area and aspect ratio are performed by morphological processing and the final road extraction results are obtained.(2)Research on road centerline extraction algorithm based on the improved FCM and morphological processing.Firstly,the improved FCM was used to segment the image in this paper.Then,the mathematical morphological method was used to remove the small noise outside the road,and the morphological thinning was used extract the centerline.Finally,the final centerline extraction result was obtained through the centerline burr remove.The experimental results show that,compared with the traditional centerline extraction algorithm,the proposed centerline extraction algorithm has an average integrality of 97.98%,an average accuracy of 95.36%,and an average quality of 93.59%.The improved FCM clustering algorithm used in this paper greatly improves the anti-noise ability of the traditional FCM algorithm,and the using of the morphological thinning method reduces the calculation of the road centerline extraction.(3)Research on road reconstruction algorithm based on multiscale segmentation and distance field calculation.Firstly,three multi-scale segmentation algorithms are compared and analyzed,from which the optimal multi-scale segmentation algorithm is selected to segment the image and the initial road area is extracted by feature extraction.Then,FMM(Fast Marching Methods algorithm)was used to obtain the boundary distance field and the source distance field,and the minimum-cost path algorithm was used to get the initial centerline.Meanwhile,road width of each initial centerline was calculated by combining the boundary distance field.Then,a tensor field was used to connected to the broken centerline to get the final centerline.Finally,the final centerline was matched with its corresponding path width,and the final road is reconstructed.Experimental results show that the average integrity,the average accuracy and the average quality of the centerline result is 94.81%,97.49% and 92.66%,respectively.The proposed road extraction algorithm is a good way to improve the accuracy of the centerline in the previous road extraction process,the problem of fracture,the reconstruction of the road is well maintained the integrity of the road.