3D Simulation Scene Optimization Model of High-Speed Railway Virtual Reality Based on Artificial Neural Network

Applying VR (Virtual Reality) technology to the simulation of high-speed railway can undoubtedly meet people's new needs for high-speed railway cognition. In order to improve the drawing speed. In this paper, ANN (Artificial Neural network) is used to optimize the 3D simulation scene of high-speed railway virtual reality. A feature point processing optimization algorithm based on BPNN (BP neural network) and FCN (Fully Convolutional Network) neural networks was proposed Based on SIFT (Scale-Invariant Feature Transform) framework, this algorithm uses pre-trained FCN neural network to detect the image and determine the pixel range of the main target, thus improving the pertinence of feature point detection and avoiding background noise interference. The research results show that this method can effectively eliminate mismatches and improve the accuracy of the results, and its average time consumption is reduced by 24.03% compared with CNN algorithm. ANN greatly optimizes the scene and improves the rendering effect and running efficiency.