Early Detection And Analysis Of Maize Weevil Based On Soft X-ray And Low-field Nuclear Magnetic Resonance

Sitophilus zeamais is a typical hidden insect of grain.Early detection of S.zeamais is of great significance to reduce the loss of stored grain and improve the quality of stored grain.In this paper,we used soft X-ray and low-field nuclear magnetic resonance?LF-NMR?to detect S.zeamais.According to the image information taken by soft X-ray,observing the whole growth period of S zeamais in wheat.The gray histogram and texture features of the image were extracted,and the discriminant accuracy of S.zeamais at different infested stages?especially at early infested stage?was judged by different algorithms.Leading the LF-NMR method into the stored grain pest detection field.The characteristic relaxation parameters of wheat with different moisture content and S.zeamais in different stages were measured,which provided theoretical basis for the detection of hidden insects by LF-NMR.Moreover,soft X-ray assisted LF-NMR was used to analyze the growth and development information of S.zeamais in wheat with different moisture content and simulate the growth of S.zeamais in wheat with different insect states,so as to realize the qualitative and quantitative detection of the early occurrence of hidden insects in grain.The main conclusions are as follows:1.According to the image information of wheat taken by soft X-ray,the gray histogram and texture features were extracted.Linear discriminant analysis?LDA?and quadratic discriminant analysis?QDA?were used to classify and identify infested wheat.Under 95%confidence interval,the accuracy of identification increased with the increase of the volume and maturity of S.zeamais in three different moisture content of wheat?12%,15%and 18%?.When the uninfested wheat was compared with those infested by S.zeamais eggs,larvae,pupae,adults and hollowed-out wheat,the discriminant accuracy of LDA was above 95%,and the discriminant accuracy of QDA was above 90%.The classification accuracy of LDA and QDA discriminant models increased gradually from early infested stage to late infested stage of S.zeamais.The average discriminant accuracy of early infested stage was 65%,and the highest accuracy rate was 78%.The relatively high accuracy showed that soft X-ray could be used for early detection of hidden insects.Partial least squares?PLS?method was used to predict the relationship between the extracted characteristic parameters?gray histogram and texture features?and the length of wormholes,the width of wormholes and the growth length of S.zeamais.The determination coefficient R² was above 0.625?P<0.001?.2.The results of LF-NMR showed that the characteristic relaxation signals of wheat and S.zeamais were different.The range of characteristic relaxation time of wheat with different moisture content was 0.50².02 ms,and the range of characteristic relaxation time of four different insect stages of S.zeamais was 37⁸7ms.By comparing the signal values of 22.03%moisture content per unit mass of wheat with different insect status of S.zeamais,it was found that there were significant differences between them?P<0.05?,which laid a foundation for analyzing whether there was hidden insect S.zeamais in normal stored wheat?moisture content is 12%¹8%?.Combining the total signal intensity of nuclear magnetic resonance with the value of nuclear magnetic resonance parameters,LDA was used to identify the adults of S.zeamais mixed in wheat,the accuracy rate was 85.0%.It showed that this method could classify and discriminate the number of adults of S.zeamais mixed in wheat with different moisture content.3.The growth and development of S.zeamais in wheat with different moisture content?12%¹8%?were analyzed by LF-NMR.The characteristic relaxation time of wheat and S.zeamais could be detected,the characteristic relaxation time of wheat was between 0.43¹.00 ms,and that of S.zeamais in wheat was between 37.65⁸6.97ms.The difference of the relaxation time range between the infested wheat and the uninfested wheat was significant,especially after 18 days of infestation.According to the change of the ratio characteristic peak area of S.zeamais and wheat,it could be judged whether wheat was infested by S.zeamais and which insect state it was in.Nuclear magnetic resonance signal parameters(T₂₁,T₂₂,P₂₁1 and P₂₂)could distinguish whether wheat was infested or not.Except for the stage of egg infestation,larvae?12days?,pupae and adults of S.zeamais infested wheat could be distinguished from uninfested wheat by LDA algorithm,and the correct rate of discrimination was87.8%.Therefore,LF-NMR could be used for early detection of S.zeamais infested with wheat.4.Conducting the mixed simulation experiment of S.zeamais in the external of different moisture content wheat and the actual simulation experiment of S.zeamais growing and developing in different moisture content wheat.Establishing a mathematical model between different insect states of S.zeamais grew in or out of wheat and the insect peak signal proportion P₂₂.It was found that there was a good correlation between them,and the determination coefficient R² was above 0.9,which provided a new method for early and accurate detection of hidden insects in practical production.