Abstract: Garlic planting needs to meet the agronomic requirements of scale-bud up and vertical planting, so the orientation of garlic seed is the key to realize garlic mechanization planting. The orientation technology of garlic seed based on machine vision is an effective method to solve the orientation problem of general garlic varieties. Relevant research focuses on the algorithm of image recognition of garlic seed tips and the control orientation technology after recognition. At present, there are a variety of algorithms for garlic scale-bud image recognition, and the recognition accuracy is up to more than 99%. The static garlic seed image collection is mainly taken as the research object. Considering the impact of moving garlic seeds, light, dust, etc. In the actual working environment on the collection of garlic images, there are not many studies on analysis of the recognition results of garlic images collected in different orientations, the impact of shape feature differences between garlic species on recognition results and comprehensive identification of garlic bud tip positions by collecting images of garlic seeds from multiple angles.  At present, the research on the control orientation technology after image recognition shows that the motion of the orientation mechanism is intermittent and executed after image recognition. The efficiency of garlic seed orientation operation is low and it is difficult to adapt to the operation speed of garlic planter. So far, there has been no garlic seed intelligent directional technology based on machine vision garlic seed seeder product. The practical gap of existing technologies is mainly reflected in the insufficient integration of the key technologies of single seed selection, image recognition, orientation mechanism and control. Therefore, with the innovative research of the organic combination of the three technologies, the scale-bud identification and orientation are parallel. The orientation of garlic seeds was achieved under the state of continuous motion. The orientation efficiency of garlic seeds was improved while the scale-bud recognition algorithm and the motion of the orientation mechanism were simplified. This paper can provide reference for the practical application of this technology, so as to promote the mechanization and intelligent development of garlic production.

Key words: Garlic cultivation, Machine vision, Garlic orientation, Identification algorithm of garlic seed scale bud