Research On Examination Interior Quality Of Tomatoes With Near Infrared Spectroscopy

The tomato not only tastes delicious, is of nutrition richly, also has certain medicinal value, is one of fruits and vegetables that people like in daily life. The content of soluble solid, total sugar, total acid in tomato has a great impact on tomato's flavor quality and taste, and the Vitamin C content in the tomato often is identified as an important indicator of the quality of their products and storability. At present it is urgent to understand and grasp how many the content of soluble solid, total sugar, total acid, VC in the tomato in its quality breeding, nutrition appraisal, food processing and storage and transportation field. The traditional chemical analysis methods belong to destructive measurement, and exist shortcomings of experiment trouble, time-consuming and laborious, and expensive drugs with experiment could easily lead to pollute environment, and so on. So it is of great scientific significance and practical value to explore green, rapid, nondestructive method to determine tomato interior quality based on the near-infrared spectroscopy technology.Based on Near Infrared Spectroscopy technology and wavelet analysis theory, the research of nondestructive methods about testing soluble solids, total sugar, total acid and vitamin C content in tomato was studied by using of OPUS software and MATLAB software. The main conclusions are as follows:1. The multiposition spot average spectrum and the single position primitive spectrum were studied in order to establish the tomato interior quality near-infrared PLS calibration models, the result was that the calibration model of the multiposition spot average spectrum was better. Simultaneously comparing the influence of the average spectrum and the primitive spectrum on the forecast performance of calibration model finally indicated that the average spectrum can improve the accuracy of the calibration models about nondestructive determination of ingredient in tomato effectively.2. In order to search an appropriate pretreatment method to nondestructive measure the interior quality of tomato, the effect of 10 conventional pretreatment methods on PLS calibration model has been compared, it can be obtained that the constant offset elimination was the best effect on establishing tomato soluble solids and the total sugar quantitative analysis model; the min-max normalization was an efficient method to nondestructive measure to Vitamin C in tomato with near-infrared spectroscopy. The quantitative analysis models were established by application of partial least squares, in order to measure interior quality of tomato nondestructively by near infrared spectroscopy based on the optimization pretreatment methods.3. The near infrared spectroscopy of tomatoes was denoised based on wavelet analysis to build the model of vitamin c. On the same scale decomposition dbN wavelet function has been optimized, study found that db9 was the best denoising wavelet function. Different decomposition scale of db9 has been optimized, the observations suggested, 6-scale decomposition denoising PLS calibration model was the best, the correlation coefficient R between predictive value and actual value was 0.897, RMSECV was 1.0700mg/100g, RMSEC was 1.0800mg/100g, RMSEP was 1.170 mg/100g.4. The appropriate method to nondestructive measure the content of SSC, TS, TA and VC has been searched based on wavelet denoising and near infrared spectroscopy analysis technology, finally the models were built on optimization band, The various models' correlation coefficient R was in turn 0.958, 0.930, 0.875, 0.897, RMSECV was in turn 0.307%, 0.466%, 0.0360%, 1.0700mg/100g, RMSEC was 0.309%, 0.469%, 0.0363%, 1.0800 mg/100g in turn, RMSEP was in turn 0.366%, 0.255%, 0.0386% and 1.170mg/100g, Which indicated forecast precision of the quantitative analysis models was high and the stability of them was good.5. It has been compared the effect of the different spectrum regions on forecast performance of tomato calibration model about SSC, TS, TA and VC separately. The result indicated that it was not notability about the influence of different spectrum area to each model's forecast precision.