基于深度学习的地图线性要素走向判定方法OACSTPCD

Map inspection,as a means of erroneous map discovery,is still based on labor-intensive visual inspection.The prominent issues by this means,such as high professional requirements for inspectors,high labor intensity,low efficiency of map inspection,and lacking of diverse in-spection rules,should be solved properly.In this paper,we used a map linear element extraction and direction determination method based on deep learning to inspect the direction of Ningxia Hui Autonomous Region boundary and that of Qingshui River in the administrative map of Ningxia Hui Autonomous Region,and used metrics to evaluate the accuracy of inspection results.According to the results,the minimum value of pixel accuracy(PA)is 91.86%for the determination of boundary elements of Ningxia Hui Autonomous Region,and the minimum value of Recall is 85.42%for the determination of linear elements of Qingshui River.The experimental results show that this method can support customizable linear element orientation rules,and provide a solution with high accuracy and efficiency to the issues mentioned above in the labor-intensive map inspection.