Abstract
Visual odometry (VO) is a method used to estimate the spatial movement of vehicles or camera-equipped systems by analyzing visual data from the environment. It is a cost-effective and accurate alternative to systems like Global Navigation Satellite Systems (GNSS) and Inertial Navigation System (INS), and is particularly favored in autonomous vehicles, unmanned aerial vehicles (UAV s), and augmented reality applications. This study focuses on the performance challenges faced by visual odometry systems in low-light environments, where traditional systems may not be effective and presents a novel approach, Contrast-Adaptive Block Optimization (CABO), to enhance VO in low-light environments. CABO divides images into blocks and selectively applies Contrast Limited Adaptive Histogram Equal-ization (CLAHE) to darker regions, optimizing image quality without compromising natural appearance. By improving feature extraction and matching, CABO enhances VO accuracy and robustness. We utilized the Karlsruhe Institute of Technology and Toyota Technological Institute (KITTI) dataset, we demonstrate the effectiveness of the proposed method in improving position and orientation estimation under challenging lighting conditions.