Basic Research On Application Of Terahertz Spectrum And Image Thin Shell Seed Quality Detection

The seed is the "chip" of agriculture,and its quality is a prerequisite for high-quality and increased yield of crops.In seed quality parameters,water content and plumpness are closely related to seed vigor.Fresh and full seeds are usually rich in water,full and active.However,seed kernels of thin-shell crops are wrapped in seed shells,which do not destroy the characters of seed shells that are difficult to obtain seed kernels.Terahertz(THz)radiation is located between microwave and infrared light,which has the advantages of both.The THz spectral region near microwave has the characteristics of low energy,good perspective and no ionizing radiation damage.And near infrared spectral region has fingerprint spectral characteristics.Therefore,the THz wave can image its interior through the seed shell,and obtain the THz spectrum and image information of the seed kernel.It can describe the traits of seed kernels,and can be used as a better solution for the quality inspection of shell crops.(1)Ginkgo is a typical recalcitrant seed,which is very easy to lose water and harden and lose its activity.It just uses the strong absorption characteristics of water to detect the moisture of THz wave.The experiment collected THz spectra and images of intact Ginkgo biloba seeds and seed slices with different moisture content.By comparing the time-domain signal and absorption coefficient of seeds with different moisture content,the influence of moisture on THz waves was analyzed.The study found that the higher the seed moisture content,the lower the time domain signal amplitude and the greater the time delay;the same moisture absorbs THz radiation strongly,and the more moisture,the stronger the spectral absorption.The law of the influence of moisture on the spectrum could provide a reference for the prediction and evaluation of moisture.From the THz images of intact seeds and seed slices,you could observe the image characteristics of the influence of moisture.As the moisture decreased,the shrinkage of the kernel appeared inside the seeds.There was a certain linear law between the shrinkage and the moisture,and there was a law of distinction with between light and dark images of THz images of different moisture.Using the seed shrinkage area ratio and the average gray value of the ROI image as reference indicators,a linear regression equation for moisture evaluation was established,which could provide a reference plan for seed moisture prediction.(2)The low-energy,perspective and fingerprint characteristics of THz wave were used to carry out the detection calculation of seed plumpness and the research on the detection method of internal defects of thin-shell sunflower seeds.A THz time-domain transmission imaging system was used to collect THz transmission images of insect infestation,defects and intact sunflower seeds.The average spectra of the regions of interest of the seed kernels and seed hulls were extracted respectively,and their time-domain signals and absorption coefficients were compared and analyzed.It was found that the absorption of THz waves by the kernels in the range of 0.5-2.0 THz was significantly stronger than that of the seed hulls.The characteristic image corresponding to the absorption coefficient of this frequency band and the contrast images of the three frequency bands of 0.5-1.0THz,1.0-1.5THz and1.5-2.0THz are extracted.After the image was segmented by the threshold,the kernel-to-shell area ratios of the THz image and the visible light image were calculated.Among them,there was a significant difference between the fullness of defects(worm infestation,defect)and intact seeds,and the intact seeds could be correctly identified by using 23% as the threshold.At the same time,using the fullness calculated from the visible light image of the sunflower seeds after peeling as a reference value,a linear fit was performed between the THz image and the fullness of visible light image,and a THz transmission image measurement model for the fullness of sunflower seeds was established.Using sunflower seed samples that were not involved in the modeling to evaluate the predictive ability of the model,the model’s prediction coefficient of determination was 0.91,and the prediction root mean square error was 4%.The research results show that the non-destructive detection of sunflower seed defects and fullness based on the time-domain THz transmission map is feasible.The thesis provides a new solution for the quality evaluation of shell seeds,and provides a reference for the non-destructive testing of other dry,light and thin packaging materials.