Machine vision is an important method for navigation of agriculture machine. A machine vision algorithm is presented to obtain a guidance directrix for an automatic machine guidance system. Based on the operating environment characteristics of the farmland,this method focuses on a gray-scale transform by using color factor (2G-R-B),and a binary transform by using OTSU threshold method,and a known point improved Hough transform to extract the guidance directrix. This method can solve the problem of low speed of traditional Hough transform. Experiments show that it takes only 100 ms on average to deal with a 640 pixel×480 pixel color image by using the algorithm and correct recognition rate reaches to 92%. It can meet the requirements of real-time processing. It is a fast and effective algorithm for image processing,and is of some practical value.