In order to obtain the quality of peanut seedling rapidly and accurately, a method based on machine version was put forward to evaluate the quality of peanut seedling. Firstly, a field walking robot was developed which can ensure the robot accurate moving automatically and keep a constant speed. The peanut image information was achieved by the camera configured on the robot, and the picture coordinate information was recorded by global position system. The number of peanut seedlings, canopy projection area of peanut seedlings and the coordinate position of peanut root was achieved based on machine vision. Secondly, the evaluation index of seedling quality was purposed, including the peanut seedling deficiency rate and peanut vitality index. The peanut seedling deficiency rate was calculated by the number of peanut seedlings and the coordinate position of peanut root, and the peanut vitality index was computed by the canopy projection area of peanut seedlings. In order to obtain the peanut number and its canopy projection area, a fast and accurate recognition method of peanut based on image adaptive classification algorithm was purposed. Peanut seedling extraction operator was proposed to enhance the robustness, and the K-means clustering method was used to automatically determine the optimal threshold for image segmentation, which avoided the environment disturbance and separated the peanut plants correctly. Then by using the global image segmentation combined regional image segmentation, the single peanut seeding was separated for farmland. Finally, the envelop area and its center position coordinates of each peanut seeding were obtained through image detection technology. Through data validation, the average recognition rate reached 95.4%, which indicated that the algorithm was feasible. Compared with the manual test, the average error of peanut seedling spacing was 5.35mm, and the correlation of peanut seedling deficiency was 0.991 (Pearson correlation coefficient). There was high consistency between manual and machine vision evaluation.