From Traditional Methods to Artificial Intelligence: A Review of Non-Line-of-Sight Analysis Techniques

Abstract

This study explores the current state, core methodologies, and major challenges associated with non-line-of-sight (NLOS) sensing technologies. NLOS sensing enables the detection of objects and individuals outside the direct field of view and has critical applications in disaster response, security, and autonomous systems. Despite its growing potential, the field faces technical limitations, including restricted resolution, complex data processing, and multipath propagation effects. A wide range of approaches is examined, including both active and passive systems, SPAD and CCD/CMOS sensors, confocal and non-confocal imaging techniques, acoustic methods, and artificial intelligence-based models. The study also emphasizes innovative experimental setups and complex scene designs to evaluate system performance under realistic and challenging conditions. Furthermore, diverse evaluation metrics are discussed to support both numerical and perceptual analysis of system outputs. In conclusion, NLOS sensing is a complex field that requires an interdisciplinary approach, but it holds great potential for the scientific community and practitioners due to the opportunities it offers. This study has contributed to the current body of knowledge and provided suggestions that will guide future research. © 2026, Sakarya University. All rights reserved.