Abstract: With the advancement of intelligent mine construction, machine vision technology has been increasingly applied in coal mine production, particularly demonstrating great potential in the intelligent perception of roadway headings. We systematically reviews the key role of vision sensors in fully mechanized coal mining faces, elaborating on the principles, advantages, disadvantages, and typical application scenarios of major types such as monocular, binocular, multi-camera systems, and structured light cameras. Furthermore, focusing on roadway headings, it discusses the technical pathways and key steps of machine vision in pose detection of the roadheader body and cutting head, covering image acquisition schemes, feature recognition, pose calculation methods, and system deployment strategies. Research indicates that intelligent perception technology based on machine vision can effectively overcome the bottlenecks of traditional methods in terms of real-time performance, accuracy, and stability, providing strong support for unmanned, safe, and efficient coal mining.