Conventional Camera Performance in Dark Environments – Noise Reduction and Visibility Enhancement

Capturing images in dark conditions is inherently difficult due to limited photon counts, high sensor noise, and di- minished contrast. This paper explores how conventional methods and modern deep learning approaches address these challenges. We review classical enhancement algorithms alongside advanced models such as CNNs, Transformers, GANs, and diffusion-based methods. Furthermore, recent hybrid paradigms combining event-driven sensing, physics-informed priors, and multimodal integration are analyzed. Comparative experiments on public low-light datasets reveal key trade- offs between noise reduction, texture preservation, perceptual realism, and efficiency. The study outlines implications for practical domainsincluding surveillance, healthcare imaging, robotics, and photography.