In order to ensure recognizing crop row automatically with some robustness against noise environment and to confirm its possible intelligence for an agriculture mobile robot working in the fields, a crop row recognition method was presented by using genetic algorithm with surface-strip model to detect the crop row imaged in the gray scale image without any preprocessing. The accuracy and stability of the proposed visual recognition of crop raw with high robustness against noise such as sunlight condition varieties and obstacles were demonstrated by artificial image and real image scanning. The robustness of the method against environmental noises and the effectiveness of the method for real-time recognition have been verified by using real rural images.