The extraction of positional features from the longitudinal lap edge straight line of the conveyor belt in a pipe belt conveyor holds significant importance in the detection of conveyor belt torsion faults. This study presents a proposed algorithm for detecting the longitudinal lap edge on a conveyor belt using machine vision. The algorithm involves several steps, including extracting regions of interest (ROI) from conveyor belt images and converting them to grayscale. Local enhancement is achieved through image stretching, and edge features are extracted using the Prewitt operator. Morphological methods are then applied to address impurities, and the lap edge is detected through straight line detection using the Hough transform. The conducted experiments involving the utilization of OpenCV on authentic images of a cylindrical pipe belt conveyor within an industrial setting demonstrate the algorithm’s proficient ability to accurately recognize the longitudinal lap edges of conveyor belts.