Large scale agricultural production at home and abroad have shown a clear trend, human control alone is unable to cope with the pressure caused by large-scale agricultural production.Both bumps in the road and drivers' fatigue driving makes the tractor deviating from the driving path, so does the agricultural implement connecting behind the tractor.Which result in damage to the crop.This paper design and developed a positioning module for agricultural implement based on image processing.The module can detect the position of crop rows automatically, and adjust the implement position to keep a certain distance between them, so as to avoid damage to crop.This paper focuses on the theoretical studies of image preprocessing, crop row detection and coordinate transformation.On the basis of light changing adaptability and algorithm complexity consideration,this paper proposes using 2Cg-Cr-Cb color feature while image preprocessing instead of 2G-R-B color feature.which turns out better light adaptability and image segmentation effect than the 2G-R-B color feature.On the hand of crop row detection, this paper optimized the traditional linear scanning method with the help of processes like projection in layers, mean filtering, denoising,wave combination etc.Possible crop areas will be picked after these steps, and it is easy to determine the exact crop row area using the CRAI valuation model introduced in the paper.These processes contributed to the reduce of linear scanning areas, and filtered the weeds to improve the efficiency of the algorithm.Besides, the paper presents an available solution with hardware and software.The module is developed with an ARM based embedded platform, Logitech C270 HD Web Camera and Sinan M600 GPS receiver.The software environment is based on embedded Linux OS, and developed with QT application framework.Results of the trial indicated the 2Cg-Cr-Cb color feature shows better light changing adaptability and image segmentation effect comparing to 2G-R-B.and it saves almost 90% time consuming.The optimized linear scanning algorithm reduce time consuming greatly and can detect the crop row accurately.The module is cheap, effective and stable, which can mect the requirements of field work.