Font Size: a A A

Seed Vigor Classification And Prediction Based On Laser Speckle Technology

Posted on:2019-07-25Degree:MasterType:Thesis
Country:ChinaCandidate:Y Q ZhaoFull Text:PDF
GTID:2393330575497702Subject:Mechanical and electrical engineering
Abstract/Summary:PDF Full Text Request
Germination rate,growth potential are closely related to Seed vigor,directly determines the harvest of crops or not.Sowing seeds with low activity can cause huge economic losses.Therefore,judging seed vigor in advance has an important meaning in actual production.Laser speckle phenomenon refers to the laser speckle phenomenon produced in the biological surface,Its change is related to the activity of certain substances inside the organism,Therefore,relative to the traditional test breeding experiment to determine seed vigor,or physical and chemical analysis of seeds to determine the vitality and other methods,The method of seed vigor detection using speckle images of seeds not only does not have to wait for seed germination,but also no need to cause seed damage.In addition,the use of substances within the seed to test vitality also makes the results more convincing.The main research contents are as follows:(1)In this paper,speckle equipment for image acquisition,and self-built database.Tasks included the investigation of the germination conditions for the seeds of Pisum sativum Linn and Quercus variabilis,Aging classification and strict breeding.Collect the image information of the first day,wait for the seeds to germinate and measure the bud length,as the real seed vigor standard.(2)The collected speckle images are processed to obtain the speckle value curve representing the seed vigor.This is mainly divided into two parts:Remove the noise in the image,adjust the size of the image location,etc.,including filtering,smoothing,edge extraction,etc.A more important part is to extract each individual speckled seed image from the culture dish,which includes matrix operations,bubble sorts,etc.(3)Using caffe deep learning framework to train GD images representing the degree of variation of seed speckles,obtain seed vigor model.Achieve the purpose of using the data of hours before germination to predict the seed vigor level,the ultimate accuracy can up to 0.90.
Keywords/Search Tags:Seed vigor, Laser speckle, Deep learning, Non-destructive testing, Classification
PDF Full Text Request
Related items