Research on the Network Map Service Technology of Remote Sensing Image Intelligent Conversion Based on Gan Model

Abstract

Abstract. Based on an improved generative adversarial networks algorithm (CGAN), this paper explores a technical way to realize map transformation through autonomous learning and training of remote sensing images. Just skip the trial process vector data update and cumbersome process of mapping the basic map elements can be automatically transform, the image on the main streets and typical rules of construction material, can achieve automatic identification and transformation, greatly shorten the tile map production and update cycle, improve the efficiency of the network map service quality. The results of the test platform have proved that it can be applied to a certain extent and can basically meet the requirements of network map production.