The invention discloses a forklift obstacle avoidance method based on vision, which is used for solving the problems of high forklift obstacle avoidance cost and narrow detection range in the prior art, and comprises the following steps: S1, acquiring continuous image frames of images acquired by a forklift camera; S2, respectively acquiring scale features of corresponding objects in adjacent image frames; S3, in combination with the obtained scale features of the corresponding objects in the adjacent image frames, calculating the distance from the object in each image frame to the camera; S4,judging whether the distance between the corresponding object in the subsequent frame in the adjacent image frames and the camera is smaller than the distance between the corresponding object in theprevious frame and the camera or not; If yes, whether the distance between the corresponding object in the subsequent frame and the camera is smaller than a preset obstacle avoidance threshold value or not is judged; And if the obstacle avoidance threshold value is smaller than the preset obstacle avoidance threshold value, stopping forklift operation. According to the method, image operation is carried out on the basis of an original camera, no extra sensor needs to be added, the cost is reduced, and compared with a collision switch and an infrared sensor in the prior art, the detection rangeis larger.