Study On The Machine Vision And Olfactory Visualization For Fruit Maturity Detecting

China is a large country of fruit production,but the low quality of fruit restricts the export of fruit.Improving level of the nondestructive detection technology for fruit quality can enhance the competitiveness of China’s fruit industry.In recent years,nondestructive detection technology is widely used in the detection of agricultural products.This research attempted to apply machine vision and olfactory visualization technology to the detection of fruit maturity during storage,by taking bananas and mangoes as samples.And expected to obtain new rapid nondestructive detection methods for fruit maturity based on changes of fruit surface color and volatile odor.The main contents of this research are as follows:1.Study on detection of fruit maturity based on machine vision1)The hardware system of machine vision which was suitable for the detection of fruit maturity was built.Taking bananas as samples,devices such as illumination source,camera,lens and image background were selected,and the light box was designed.2)The discrimination of banana maturity based on machine vision.12 color features namely (?),(?),(?),(?),(?),(?),H_d,S_d,I_d,L_d,a_d,b_dand 4 texture features namely Con,Asm,Ent,Corr were extracted from images of banana samples,and these 16 features constituted the information features of the images of bananas.Then Fisher Linear Discriminant(FLD)model and Support Vector Machine(SVM)model for evaluation of banana maturity were established.The discrimination accuracy were 77.14%and74.29%.3)Two indicators related with fruit maturity were detected.Taking bananas as samples,features extracted from images of bananas were used as input,and the value of fruit hardness and the total soluble solid were used as the output respectively,the support vector machine regression prediction model was established.The results showed that,the root means square error(RMSEP)and the correlation coefficient(R_P)for prediction set were respectively 0.3206 and 0.80173;The RMSEP and Rp were respectively 0.4775 and 0.75127 in the model for total soluble solid.2.Study on detection of fruit maturity based on olfactory visualization technology1)Selection of colorimetric sensors and fabrication of colorimetric sensor array for detection of fruit maturity.Taking bananas as samples,gas sensitive materials of colorimetric sensors were chosen.And colorimetric sensor array was set up.2)The detection of banana maturity based on olfactory visualization technology.48 features were extracted from each colorimetric array.Principal component analysis was carried out.And the first 13 principal components were used as the input and the maturity level as the output.FLD,Bayes and SVM model were established to discriminate banana maturity.The results showed that,the discrimination accuracy of banana maturity got the highest value when reaction time was 10 minutes between colorimetric array and the gas from banana samples.Bayes classification got better result,the discrimination accuracy for the training set and the prediction set were 84.76%and 77.14%respectively.3)Quantitative prediction of hardness and total soluble solid of fruit based on olfactory visualization technology.The first 13 principal components were used as the input and value of hardness and total soluble solid as the output,support vector machine regression prediction model was established.The regression coefficients are above 0.6,and olfactory visualization technology is not excellent for the prediction of hardness and total soluble solid of fruit relative to machine vision.3.Study on detection of fruit maturity based on the fusion of machine vision and olfactory visualization technology.1)Study on fusion method of fruit maturity detection.The method of feature level fusion based on machine vision and olfactory visualization was established.The SVM model for discrimination was established.The discrimination accuracy for training set and prediction set were 97.14%and 85.86%,respectively.It was shown that the detection method of banana maturity based on fusion technology was feasible.2)Quantitative prediction of hardness and total soluble solid based on fusion information.Taking bananas as samples,the fusion data was used as the input and value of hardness and total soluble solid as the output.The support vector machine regression prediction model was established.The prediction correlation coefficient between fusion data and indicators that included hardness and total soluble solid were both above 0.8.4.Verification of quantitative prediction models for fruit maturity,hardness and total soluble solid.Taking mangoes as samples,the quantitative prediction models which were established using banana samples for evaluation of samples’maturity,hardness and total soluble solid were tested.The experimental results showed that the maturity accuracy rate was 98.71%for the training set and 87.50%for the prediction set.The correlation coefficients between maturity and indicators that included hardness and total soluble solid were more than 0.82.The quantitative forecasting models for fruit maturity detection,hardness and total soluble established by the example of banana,showed consistent correlation in the mango test;the prediction trend of the model was the same,the specific parameters of the model were corresponding to the different fruit products.For mango,the model achieved better detection results,which indicates that the established method for fruit maturity detection based on machine vision and olfactory visualization technology is robust.The results showed that the machine vision and olfactory visualization technology is feasible for the detection of fruit maturity,and the fusion technology has higher detection accuracy than any of the single detection technology involved.The results of this study can be extended to the same or similar varieties of fruits.