Abstract: Super hybrid rice is a kind of rice variety that provides excellent quality and high yield. Super hybrid rice is very popular as a large-scale rice crop in China. The precision seeding of tray nursing seedling requires (2±1) seeds per hole. At present, the performance of seed metering devices can reach 1-4 seeds per hole, but the rate of single seed is high (more than 20%), and there are cavities. Due to the factors such as the germination rate, survival rate, characteristics of blanket and injured seedling rate, the seeding results cannot reach 1-2 plants per hole. Therefore, seeding target should be raised to 2-3 seeds per hole, which will reduce single-grain rate and eliminate the presence of cavities. When the seedlings for planting rice, vegetables, flowers, and so on, especially the super hybrid rice were used to mechanized seeding, the situation of cavity and single seed commonly exists. In order to achieve the goal of digital and intellectual precision seeding, which is difficult to achieve by the traditional seeding device or seed metering device, focusing on imitating the manual sowing principle with the monocular camera as eyes, the vision inspection systems as the brain, and the hill-drop device performing hand movements, this paper presents a new method of intelligent reseeding based on machine vision technology, and develops an intelligent reseeding devices, which is mainly used for quality detection and compensating seeds of the super hybrid rice. Firstly, the image of seedling tray was collected by CCD (charge coupled device) camera, the positions of cavities and the holes of single seed were obtained by image processing and analysis, and then the positioning mechanism and reseeding mechanism were used to realize the function of picking up the seed from seed groove and dynamically reseeding on the designated location. Applying LabVIEW graphical programming software, on-line testing and motion control system of seedling planting quality was developed to realize the task of intelligent reseeding. According to the statistics of the test results, the hill-drop machine can meet the production requirements of 450 trays per hour and the average reseeding time per hole was about 2.48 s when the reseeding rate was less than 2%. The average processing time of an image was about 0.518 s, and the accuracy was more than 95%. The paper achieves the intelligent reseeding according to the required number of seeds per hole. As a new precision seeding technology, the method will raise the level of planting accuracy, and have important scientific significance and application prospects. However, in the practical application of intelligent reseeding, considering the single dibbler's working efficiency is still relatively low, the integrated design method can be used with 2-3 seeding devices according to reseeding rate and productivity, or the lightweight, fast reseeding mechanism, such as SCARA, parallel or serial manipulator, which can play a better advantage of intelligent reseeding. In addition, with the rapid development of robot technology, the cost of intelligent dibbler will become lower and lower, and the efficiency will be higher and higher.