SLAM 3D Digital Terrain Mapping with SqueezeNet Driven by Road Traffic Data

Abstract

In order to improve the efficiency of dynamic visualization of large-scale road traffic data in a web environment, this paper proposes a SLAM 3D digital terrain map visualization method using SqueezeNet in a web environment. We propose a hierarchical organization method of the road network, taking into account road attributes and drawing roads at different view heights; a multiroad merging method based on the line segment indexing feature of WebGL (web graphics library) technology, and optimizing the scene by combining view rejection and multithreading technology. The prototype system was developed and case studies were carried out using the national road network data as an example. The experimental results show that the frame rate of large-scale road traffic data visualization in the network environment is above 40 frames/second, which is 20–30 frames/second higher than that of Baidu’s ECharts GL visualization method.