Port target recognition based on high resolution images
Based on the analysis of the spatial relationship characteristics of breakwater in remote sensing images and its related prior knowledge,this paper conducts nearshore target detection from three aspects:sea-land segmentation,breakwater extraction and recognition.Firstly,preprocessing is applied to the high-scoring No.2 image,and sea-land segmentation is performed on the binary image of the processed image using an improved region growing method.Secondly,the breakwater is extracted from the coastline according to the prior knowledge and spatial relationship characteristics of the breakwater in remote sensing images.Finally,the breakwater is identified by Hough transform.Experiments are conducted to extract two types of breakwaters,with an average overall accuracy of 96.41%and an average Kappa coefficient of 0.9603.The results show that this method can accurately,stably,and efficiently achieve automatic recognition of breakwater.
NETL
NSTL
万方数据
