A Review of High-Precision Ranging Methods Based on Machine Vision

High-precision ranging technology utilising machine vision is a pivotal enabler for fields including robot navigation, autonomous driving, and advanced manufacturing. This paper presents a systematic overview of established machine vision ranging techniques, with particular emphasis on the fundamental principles, performance attributes, and application areas of three methodologies: binocular stereo vision, structured light, and Time-of-Flight (ToF). The paper analyses and contrasts the strengths and weaknesses of each approach: binocular vision, a passive method, offers low cost, but its accuracy and robustness are contingent on environmental texture; structured light achieves high-precision measurement of surfaces lacking texture via active pattern projection, demonstrating considerable advantages in short-range three-dimensional reconstruction; and ToF technology exhibits unique performance capabilities in long-range and real-time depth sensing. Furthermore, this paper explores the crucial role of deep learning in enhancing ranging accuracy and robustness and, with reference to application examples such as robot navigation, forecasts the key trends in future technological developments towards multi-sensor fusion, novel hardware components, and increased intelligent automation.