Enhancing Around View System for VEHICLES: Lut Correction Via Deep Learning

Abstract

We present around view system enhanced by deep learning techniques. The around view system is an essential component in autonomous driving and parking systems. Traditional around view systems have been designed using handcrafted algorithms. Moreover, utilizing pre-generated look-up tables (LUTs) has not been effective in reducing errors that occur during driving. We propose an algorithm that leverages deep learning-based methods to correct while driving. By using deep learning for feature matching, we can update the LUT to decrease the errors between different viewpoints. © 2024 IEEE.