Title: Image quality estimation based on visual perception using adversarial networks in autonomous vehicles

Authors: D. Vijendra Babu; A. Umasankar; K. Somasundaram; C.M. Velu; A. Sahaya Anselin Nisha; C. Karthikeyan

Addresses: Department of Electronics and Communication Engineering, Aarupadai Veedu Institute of Technology, Vinayaka Mission's Research Foundation, Deemed to be University, Paiyanoor 603104 Tamil Nadu, India ' Department of Electrical and Electronics Engineering Technology, Yanbu Industrial College, Yanbu, Kingdom of Saudi Arabia ' Institute of Information Technology, Saveetha School of Engineering, SIMATS, Chennai, India ' Saveetha School of Engineering, Saveetha Institute of Medical and Technical Sciences, Deemed to be University, Chennai, Tamil Nadu 602105, India ' Department of Electronics and Communication Engineering, Sathyabama Institute of Science and Technology, Deemed to be University, Chennai 600199, Tamil Nadu, India ' Koneru Lakshmaiah Education Foundation, Deemed to be University, Vaddeswaram, Guntur, Andhra Pradesh 522502, India

Abstract: To improve autonomous cars, the dynamic systems method is re-enacted. Due to the unreality of the sensors employed in vehicles, human creation of the surrounding environment and objects is necessitated. We propose a novel efficient method for generating accurate scenario sensor data using limited LIDAR and video data from an autonomous vehicle. A new SurfelGAN network recreates realistic camera pictures to recognise the cars and moving objects in the scenario. The suggested approach uses real-world camera image data from Waymo Open Dataset to evaluate actual scenarios for autonomous vehicle movement. A new dataset allows for simultaneous analysis of two autonomous cars. This dataset is used to test and explain the proposed SurfelGAN model. GAN is the greatest technique for capturing realistic pictures. The machine generates precise sensor data that is used to identify obstacles, cars, and other moving objects in the route of an autonomous vehicle. The autonomous car approaches the destination by recreating a surfel scene. Pictures are collected using semantic and instance segmentation masks.

Keywords: generative adversarial networks; GAN; visual perception; image quality assessment; IQA; autonomous vehicle; SurfelGAN.

DOI: 10.1504/IJESMS.2024.135117

International Journal of Engineering Systems Modelling and Simulation, 2024 Vol.15 No.1, pp.37 - 46

Received: 11 Sep 2021
Accepted: 28 Jan 2022
Published online: 01 Dec 2023 *