Research On The Key Technology Of Vegetable Grafting Seedling Classification

The quality of vegetable grafting seedlings directly affects the grafting efficiency and survival rate.At present,the selection of grafting seedlings is mainly conducted by manual sorting,which has the disadvantages of low efficiency and high labor intensity.Based on the requirement of mechanized grafting for rootstock seedlings,this study used machine vision and image processing technology to realize nondestructive testing of the characteristic parameters of pumpkin rootstock seedlings,and determined the seedling classification standard of melon rootstock.The main contents and achievements were as follows:1.Based on the morphological characteristics of seedling used for vegetable grafting during the appropriate period of grafting,an online visual detection device for seedling used for vegetable grafting was constructed.The hardware structure of the visual acquisition and processing system for digital image of melon rootstock grafting seedling was built by using the lighting module of front and background light mixture based on LED light source.2.Based on K-Means cluster analysis method,the grading criteria of rootstock seedling quality in suitable period of grafting were preliminarily studied,and the grading criteria of grafted rootstock seedling quality were determined: Grade ? seedling was 64.29mm?plant height?79.02 mm,stem diameter>3.04 mm,cotyledon width>49.27 mm,Grade ? seedling was plant height<64.29 mm,2.58mm?stem diameter?3.04 mm,cotyledon width<36.38 mm,Grade ? seedling was plant height>79.02 mm,stem width<2.58 mm,36.38mm?cotyledon width?49.27 mm.The seedlings which were not classified as Grade ?,Grade ? and Grade ? were non-standard seedlings.It provided a theoretical basis for the construction of on-line grading system of grafted rootstock seedlings.3.This paper proposed a specific identification and detection algorithm for seedling and stem parameters of melon grafting rootstock based on image processing and analysis technology.The target object was separated from the background by G channel processing,and then the cotyledon width measurement method of rootstock seedlings was completed by fitting ellipse algorithm.Non-destructive measurement of stalk height and stem diameter of rootstock seedlings was carried out by means of extracting skeleton line algorithm.4.Based on C++ and Halcon,a set of automatic identification system of seedling quality grade for vegetable grafting was developed,which was mainly responsible for the extraction of digital image of grafting rootstock seedlings and the calculation of its characteristic parameters.Combined with the established classification standard of grafting rootstock seedlings,the online nondestructive classification of seedlings was realized.