“…Traditional machine vision al-Information Technology and Control 2023/2/52 298 ways grayscales the image first [16,23,26,27], then uses an edge detection algorithm to extract image edge features [8,16,18,23,26,27], and then uses the Hough transform [8, 16-18, 23, 26, 27, 30] and other methods to obtain the description of the lanes. The researchers use color features to detect lanes [23,31,22], and some researchers use an inverse perspective to solve the difficulty of lane detection caused by the perspective phenomenon [5,9,19].…”