Study On Phenotypic Characteristics Of Maize Seeds Based On Image Processing Technology

The detection and analysis of maize seed phenotypic characteristics has become an important research content in maize industry,which has broad application prospects in breeding and agricultural production.Traditional manual measurement is slow,imprecise,time-consuming and error prone.In view of the above problems of seed phenotypic characteristics,this paper selects four different maize varieties as the research objects,and takes CCD camera to take pictures of their ears and seeds respectively.The collected seed images are preprocessed by r-component method,median filtering method,maximum inter class variance method,open operation and closed operation combination method.On this basis,the ear length,ear height and seed weight of maize ear are extracted Ear width,row number per ear,grain number per row,long axis length,short axis length,perimeter and area of corn kernel were analyzed.The results of image measurement and manual measurement were compared and analyzed.Finally,the B /S architecture was used to automatically collect the phenotypic characteristics of corn seeds through Java springboot framework.The main results are as follows:(1)The system includes user management module,data management module,data analysis module and platform basic maintenance module,which can provide reliable data and query function for breeding experts.(2)The quantitative characteristics of ear length,ear width,rows per ear and grains per row were obtained automatically,and the image analysis value was compared with the manual measurement value.The relative error of ear length varied from 1% to 3%,and the average relative error was 2.2%;The relative error of ear width varied from 1% to 5%,and the average relative error was 2.9%;The relative error of row number per spike varied from 2% to 3%,and the average relative error was 2.5%;The relative error of grain number per row varies from 2% to 4%,and the average relative error is 3.1%.(3)The results show that the relative error of corn kernel number varies from 1% to 2%,and the average relative error is 1.2%;The relative error of long axis length varies from 2% to 3%,and its average relative error is 2.5%;The relative error of short axis length varies from 2% to 5%,and its average relative error is 3.2%;The relative error of perimeter varies from 2% to 4%,and its average relative error is 3.5%;The relative error of area varies from 1% to 4%,and its average relative error is 3.8%.The average relative error of all the above measurements is less than 5%,and the error value is small.The experimental method is effective and feasible.The visualization of phenotypic characteristics data of maize seeds is realized,so as to clarify the phenotypic characteristics of these maize seeds and provide convenience for future scientific research.