Multi-scenes Image Stitching Based on Autonomous Driving
Lang Wang, Wen Yu, Bao Li
Abstract
Most of the previous image stitching methods only perform well on standard datasets. These methods assume that the images are collected under ideal conditions such as sufficient light and low noise. However, these methods can't be applied well to more complex practical scenarios. This paper proposes a multi-scenes image stitching method for autonomous driving, which uses autoencoder networks to extract feature points. The feature point extraction network includes a dimensionality-reduced feature extraction path and a precisely located symmetrical decoding path. The proposed method can provide a stable wide field of view for autonomous vehicles under different weather and different lighting conditions. Multiple experimental data show the effectiveness of the method.