Row-oriented spraying technology can improve the utilization rate of pesticides, protect the environment and reduce pesticide residues. A vision based row-oriented spray control system for field cabbage was established. The improved ExG algorithm was used to extract color information, and the method of OTSU and morphological opening and closing operation were used to segment crops and background. A method of cabbage crop row localization and multi row adaptive ROI extraction was proposed. In the ROI of strip segmentation, the feature point set was collected based on the limited threshold vertical projection, and the crop row centerline was obtained by linear fitting of the feature point set by the least square method. The offset information of crop rows was obtained based on the geometric relationship of the centerline. A row offset compensation model was established based on the kinematic characteristics of the row mechanism, and row-oriented spray control system based on PID trajectory tracking algorithm was designed. Laboratory tests showed that the accuracy of crop row recognition was 95.75%, and the average algorithm time-consuming was 77ms. Field tests showed that under different periods of illumination, the recognition algorithm had the best test results in the time periods of 09:00—11:00 and 14:00—16:00, and the average recognition deviation was kept below 2.32cm. In the weed press test, the average accuracy rate of the recognition algorithm was 95.56%, indicated that the algorithm had strong robustness.in the comparison test with other recognition algorithms, the algorithm proposed had the shortest average time consumption and the highest recognition accuracy rate, and it could be used for real-time operations.in the field row-oriented spray control system tests, the system row-oriented accuracy rate reached 93.33%, and the control algorithm could control row-oriented deviation within 1.54cm, which could meet the requirements of practical field applications.