Image-based Identification Of Rice Flower Fragrant Rice Seed

Rice is one of the most important grain food crops in the world,which is not only a traditional Chinese food,but also a staple food for more than half of the world's population.In recent years,as people's pursuit of the nutritional value and taste quality of rice has been continuously improved,non-high-quality rice seeds have appeared on the market to impersonate high-quality Daohuaxiang rice seeds,seriously damaging the health and interests of consumers.Based on the image processing technology,this paper identifies the rice cultivar Daohuaxiang No.2 by measuring the physical morphology of rice seeds.In this study,four kinds of rice seeds including,Daohuaxiang No.2,Kenjian No.5,Kenjing No.6,and Longyou No.420 in Heilongjiang Province were used as samples to calculate the physical forms of 4 kinds of rice seeds through image preprocessing and image processing software.Based on the analysis of data analysis using principal component analysis(PCA)and cosine similarity,to identify Daohuaxiang No.2.The main contents of this article are as follows:(1)In the same period of time,the four kinds of rice were collected and plastic-sealed under the same external environment.The purpose was to avoid the physical deformation of rice seeds caused by changes in the temperature and humidity of the external environment,and to ensure the accuracy of the measurement results to the greatest extent.Then the image collection of four kinds of rice seeds was performed.(2)The image processing software written by Delphi was used to preprocess the four kinds of rice seeds collected,including the RGB color image graying,gray image linear transformation,image binarization,edge extraction,and region filling.And then the center of gravity X,center of gravity Y,area,perimeter,rice seed image after pretreatment The complexity,depth,length,spheroidality,ratio of length to length,and coefficient of variation were calculated for 10 independent variables.(3)The principal component analysis method was used to analyze and reduce the data of10 independent variables.The final identification of rice cultivar Daohuaxiang No.2 was performed using the method of cosine similarity.The experiment showed that the correct identification rate of Daohuaxiang No.2 rice was 88%.(4)Based on the PHP+MySQL+Apache development of the webpage version of image processing software,through the establishment of public network,configuration variables and other operations,the development and application of the webpage version of the image processing software,and finally realize the image processing software through webpage loginoperation.Therefore,this paper discriminates the physical form of Daohuaxiang rice seeds based on image processing,using Delphi to write image processing software to preprocess the image and calculate the relevant variables.At the same time,developing the webpage version of the image processing software,and then the principal component analysis method and cosine similarity method of SPSS software were used to analyze the morphological characteristics of Daohuaxiang No.2 rice seeds and other 3 kinds of rice seeds,and finally realized the effective identification of Daohuaxiang No.2 rice seeds.