Nondestructive testing method and system for comprehensive quality of harvested bananas

Abstract

The invention discloses a nondestructive testing method and system for the comprehensive quality of harvested bananas, and the method comprises the steps: building a high-precision prediction model through obtaining a hyperspectral image, a spectral feature, an infrared radiation energy distribution image, a visible light image and a dielectric property parameter of the bananas, and carrying out the subsequent nondestructive testing of a target banana through the prediction model. The quality of the collected and cold-stored bananas is monitored in real time, the ripening grade of the banana samples in the shelf life after cold storage and ripening can be accurately evaluated, and the banana samples are classified and sorted.

Description

Technical Field

[0001] The present invention relates to the technical field of fruit detection, and in particular to a method and system for nondestructive detection of comprehensive quality of post-harvest bananas. Background Art

[0002] Bananas are typical climacteric fruits, and they are still actively metabolizing after being picked. In order to maintain the freshness and quality of bananas, they must be pre-cooled as soon as possible. After bananas are pre-cooled, they are classified and screened according to the maturity level of the fruit, which is conducive to improving the quality of the fruit.

[0003] The quality inspection of bananas after cold storage is divided into two types of quality inspection: external and internal.

[0004] External quality refers to the judgment of color, freshness, size, mechanical damage, frostbite and rot. Traditional machine vision technology is difficult to distinguish mechanical damage, frostbite, rot and freshness in the detection of external quality of fruits and vegetables due to its low precision and complex operation. Hyperspectral imaging technology can realize all-round non-destructive detection of images, with high precision and easy operation. In recent years, it has been gradually used in the detection of external quality of fruits and vegetables. For example, Xie et al. proposed a detection scheme of "Predicting banana color and hardness using new wavelength selection". It uses hyperspectral imaging technology to analyze banana color to evaluate its freshness and maturity. Freshly picked green bananas are ripened in the laboratory, and their color data is collected over time. The partial least squares (PLS) model is used to predict freshness. The prediction result of banana freshness is good, and non-contact non-destructive detection of banana color and freshness is realized.

[0005] Internal quality is an important basis for measuring its nutritional value. It is generally judged by testing the hardness, soluble solids content (SCC), titratable acid, moisture, maturity, protein and starch content of fruits and vegetables. Among them, sugar content and hardness are two important indicators reflecting the internal quality of fruits and vegetables. Sugar content can reflect the taste of fruits and vegetables, and hardness can indirectly reflect the maturity of fruits. Rajkumar et al. proposed a detection scheme for "Hyperspectral Imaging Research on Banana Fruit Quality and Maturity". They used visible near-infrared band (400-1000nm) hyperspectral imaging technology to detect the moisture content, soluble solids content and hardness of bananas under different temperature conditions, and used partial least squares analysis to process the spectral data. The principal component analysis method was used to obtain the characteristic band with the highest contribution rate in the whole band, and combined with the multiple linear regression method to establish a prediction model based on the characteristic band. The correlation coefficients of banana soluble solids content, moisture content and hardness were 0.85, 0.87 and 0.91, respectively. At the same time, it was found that the soluble solids content and hardness of bananas stored in different temperature environments showed a polynomial relationship with their maturity, while the moisture content showed a linear relationship with maturity.

[0006] However, neither of the two detection schemes proposed above has constructed a comprehensive quality index for bananas. They only test the external quality and internal quality of bananas, respectively explaining that machine vision can perform non-destructive testing on the external quality of bananas, and visible-near infrared spectroscopy and hyperspectral imaging technology can perform non-destructive testing on the internal quality of bananas. However, the prediction model only performs testing on a single quality index, and a single quality index cannot characterize the comprehensive quality of post-harvest bananas, so the detection effect and significance are not great. Summary of the invention

[0007] In view of the above problems, the present invention proposes a method and system for nondestructive detection of comprehensive quality of post-harvest bananas, which mainly solves the problem that the existing banana quality prediction model cannot characterize the comprehensive quality characteristics of the sample.

[0008] In order to solve the above technical problems, the first aspect of the present invention provides a non-destructive detection method for the comprehensive quality of post-harvest bananas, which is applied to bananas after cold storage, and comprises the following steps:

[0009] S1, collecting hyperspectral images of a number of bananas, and extracting spectral features of preset points in the hyperspectral images;

[0010] S2, collecting infrared radiation energy distribution images of a number of bananas, and converting the infrared radiation energy distribution images into visible light images;

[0011] S3, obtaining dielectric property parameters of a number of bananas;

[0012] S4, fusing the hyperspectral image, the spectral feature, the infrared radiation energy distribution image, the visible light image and the dielectric characteristic parameter at the data level to define the sample data, associating the sample data with the maturity index, building a prediction model, selecting feature variables for fusion, and defining the sample feature variables;

[0013] S5, based on the prediction model, performing decision-level fusion on the sample feature variables and the maturity index, and finally selecting batches of samples from the sample feature variables for model training, optimization and deep analysis, to complete the training task of the prediction model;

[0014] S6, re-performing the data collection tasks of S1-S3 for a single target banana, forming real-time sample data and inputting the data into the prediction model to obtain a corresponding real-time maturity index, and using the real-time maturity index as a grading condition for the maturity of the banana.

[0015] A region of interest of the hyperspectral image is selected, and spectral preprocessing is performed on the region of interest, and an average spectrum is extracted as the spectral feature.

[0016] In some embodiments, the maturity index includes at least soluble solids content, titratable acid, sugar-acid ratio and / or hardness.

[0017] In some embodiments, the infrared radiation energy distribution image is converted into a visible light image through spectral conversion and brightness enhancement, and pixel values ​​of the RGB three primary color channels of the visible light image are extracted. Then, the hue, saturation and brightness pixel values ​​of the visible light image are extracted using a color model, and the sample texture of the visible light image is extracted to estimate the surface area of ​​the banana.

[0018] In some embodiments, the dielectric characteristic parameters include complex impedance and complex dielectric constant, wherein the complex impedance is a physical constant of banana, and the complex dielectric constant is used to characterize the internal quality characteristics of banana.

[0019] In some embodiments, the process further includes S7, presetting the banana maturity level, subjecting the banana maturity level to semi-supervised deep learning and representation learning by the prediction model, performing translation-invariant classification on the input real-time sample data according to the hierarchical structure of the banana maturity level, and then transmitting the target bananas to a designated channel for packing through a sorting mechanism.

[0020] The second aspect of the present invention provides a non-destructive testing system for the comprehensive quality of post-harvest bananas, which is applied to bananas after cold storage, and comprises:

[0021] A hyperspectral camera is used to collect hyperspectral images of a plurality of bananas and extract spectral features of preset points in the hyperspectral images;

[0022] An infrared thermal imaging camera, used for collecting infrared radiation energy distribution images of a plurality of bananas, and converting the infrared radiation energy distribution images into visible light images;

[0023] An intelligent electrical parameter detection unit, used to obtain dielectric characteristic parameters of a number of bananas;

[0024] A data fusion unit, used for fusing the hyperspectral image, the spectral feature, the infrared radiation energy distribution image, the visible light image and the dielectric characteristic parameter at the data level, defining the result as sample data, associating the sample data with the maturity index, building a prediction model, selecting feature variables for fusion, and defining the result as sample feature variables;

[0025] A model training unit is used to perform decision-level fusion of the sample feature variables and the maturity index based on the prediction model, and finally select batches of samples from the sample feature variables for model training, optimization and deep analysis to complete the training task of the prediction model;

[0026] The maturity grading unit is used to re-perform the data collection task for a single target banana, form real-time sample data and input it into the prediction model to obtain a corresponding real-time maturity index, and use the real-time maturity index as a grading condition for the maturity of the banana.

[0027] The third aspect of the present invention proposes a nondestructive detection system for the comprehensive quality of post-harvest bananas, which is applied to bananas after cold storage, and includes: a first test dark box, a second test dark box, and a conveyor belt that passes through the first test dark box and the second test dark box in sequence, wherein an intelligent electrical parameter detection unit is also provided at the end of the conveyor belt, a lifting platform is provided between the first test dark box and the second test dark box, a hyperspectral camera is provided on one side of the lifting platform, and the hyperspectral camera is connected to the first test dark box through an optical fiber, wherein the first test dark box has a built-in halogen light source, and the second test dark box has a built-in annular light source, an infrared thermal imaging camera, a strip light source and an electrical parameter measurement platform, and the data output ends of the hyperspectral camera, the infrared thermal imaging camera and the electrical parameter measurement platform are connected to the data receiving end of the computer.

[0028] In some embodiments, a weighing platform is disposed at the bottom of the first test dark box.

[0029] In some embodiments, a weighing device is further provided at the end of the conveyor belt.

[0030] The beneficial effects of the present invention are as follows: by acquiring the hyperspectral image, spectral characteristics, infrared radiation energy distribution image, visible light image and dielectric property parameters of bananas, a high-precision prediction model is constructed, and the prediction model is used to perform non-destructive testing on target bananas, and the quality of bananas in cold storage after harvest is monitored in real time. The maturity level of banana samples during the shelf life after cold storage and ripening can be accurately evaluated, and they can be graded and sorted. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 This is a schematic diagram of the process of the nondestructive detection method for the comprehensive quality of post-harvest bananas disclosed in the first embodiment of the present invention;

[0032] Figure 2 This is a schematic diagram of the structure of a nondestructive testing system for comprehensive quality of post-harvest bananas disclosed in Embodiment 3 of the present invention;

[0033] Figure 3 for Figure 2 Schematic diagram of the cross-sectional structure of the inner AA area;

[0034] Figure 4 for Figure 2 Schematic diagram of the cross-sectional structure of the inner BB area;

[0035] Figure 5A schematic diagram of the spectral preprocessing result of the region of interest of the hyperspectral image in Embodiment 1 or 3 of the present invention. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solution and advantages of the present invention clearer and more specific, the content of the present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods. It is understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for the convenience of description, only the parts related to the present invention are shown in the accompanying drawings, rather than all the contents.

[0037] Embodiment 1

[0038] This embodiment proposes a nondestructive detection method for the comprehensive quality of post-harvest bananas, which is applied to bananas after cold storage. By acquiring the hyperspectral image, spectral characteristics, infrared radiation energy distribution image, visible light image and dielectric property parameters of the bananas, a high-precision prediction model is constructed. The prediction model is used to perform nondestructive testing on the target bananas in the future, and the quality of post-harvest cold-stored bananas is monitored in real time. The maturity level of banana samples during the shelf life after cold storage and ripening can be accurately evaluated, and they can be graded and sorted.

[0039] like Figure 1 As shown, the method comprises the following steps:

[0040] S1, collect hyperspectral images of several bananas and extract spectral features of preset points in the hyperspectral images.

[0041] Select the region of interest of the hyperspectral image, perform spectral preprocessing on the region of interest, and extract the average spectrum as the spectral feature, such as Figure 5 shown.

[0042] S2, collecting infrared radiation energy distribution images of a number of bananas, and converting the infrared radiation energy distribution images into visible light images;

[0043] Specifically, the near infrared rays emitted by the near infrared radiation source (such as the sun, halogen lamp) are reflected by the target object, and then the near infrared camera receives this near infrared radiation and forms the infrared radiation energy distribution image of the target object. The infrared radiation energy distribution image is converted into a visible light image after spectral conversion and brightness enhancement, so as to collect infrared images and ordinary true color images of bananas after cold storage or cold storage ripening in real time. Then, the RGB (Red, Green, Blue) three primary color channel pixel values ​​of the visible light image can be extracted by the machine vision system, and then the hue, saturation and brightness pixel values ​​of the visible light image are extracted by the color model HSV (Hue, Saturation, Value), and the sample texture of the visible light image is extracted to estimate the surface area of ​​the banana. The image is processed by threshold binarization and edge contour extraction to obtain the image gray value.

[0044] S3, obtaining dielectric property parameters of several bananas.

[0045] The dielectric characteristic parameters include complex impedance and complex dielectric constant. Complex impedance is the physical constant of banana, and complex dielectric constant (which integrates current, voltage and power factor) is used to characterize the internal quality characteristics of banana.

[0046] S4, the hyperspectral images, spectral features, infrared radiation energy distribution images, visible light images and dielectric characteristic parameters are fused at the data level and defined as sample data. The sample data is associated with mature indicators, a prediction model is constructed, and feature variables are selected for fusion and defined as sample feature variables.

[0047] In this embodiment, the data-level fusion refers to the fusion of the RGB three primary colors, color model HSV (brightness, saturation and sharpness), image texture and grayscale value, surface area, perimeter and estimated volume and other parameters of the sample's area of ​​interest with the thermal imaging temperature field, 400-1000nm hyperspectral band characteristics, and dielectric characteristic parameter characteristics including tube current and tube voltage. The above-mentioned maturity indicators include at least soluble solid content, titratable acid, sugar-acid ratio and / or hardness.

[0048] S5, based on the prediction model, performs decision-level fusion of sample feature variables and maturity indicators, and finally selects batches of samples from the sample feature variables for model training, optimization and deep analysis to complete the training task of the prediction model.

[0049] In this embodiment, the decision-level fusion refers to the use of SPA (sequential projection method) or GA (genetic algorithm) to select sample feature variables after data set fusion and maturity index (soluble solids content, titratable acid, moisture content and edible rate, etc.) indicator data for quantitative analysis of the correlation. The purpose is to reduce the dimension of the sample feature variables after data-level fusion to find data with a strong correlation with the corresponding maturity index.

[0050] The above-mentioned model training, optimization and deep analysis are aimed at the changes in soluble solid content, titratable acid, moisture content and edible rate during the storage period of the sample, which will definitely affect the comprehensive quality such as maturity. In this embodiment, the CNN-LSTM-Attention_Adaboost prediction model multivariate load prediction of sample maturity, and quality grading and sorting tasks according to maturity. Among them, the CNN-LSTM-Attention-Adaboost model predicts that the convolutional long short-term memory neural network attention mechanism is combined with the AdaBoost multivariate time series prediction (load prediction); attention mechanism + spatiotemporal feature fusion + combined model integrated learning prediction, that is, CNN-LSTM-Attention-Adaboost multivariate load prediction; CNN-LSTM-Attention-AdaBoost is a method that combines the two machine learning techniques of CNN-LSTM-Attention and AdaBoost to improve the performance and robustness of the model. Specifically, AdaBoost is an integrated learning method that combines multiple weak learners to form a strong learner, each of which is trained for different data sets and feature representations. The basic idea of ​​the CNN-LSTM-Attention-AdaBoost algorithm is to use CNN-LSTM-Attention as the base model and enhance it using the AdaBoost algorithm. Specifically, multiple CNN-LSTM-Attention models can be trained, each using a different data set and feature representation, and then their prediction results can be combined to form a more accurate and robust model.

[0051] S6, re-perform the data collection tasks of S1-S3 for a single target banana, generate real-time sample data and input it into the prediction model to obtain the corresponding real-time maturity index, and use the real-time maturity index as the grading condition for the maturity of the banana.

[0052] It also includes S7, which pre-sets the banana maturity level. After the banana maturity level is subjected to semi-supervised deep learning and representation learning by the prediction model, the input real-time sample data is classified in a translation-invariant manner according to the hierarchical structure of the banana maturity level, and then the target bananas are transmitted to the designated channel for packing through the sorting mechanism.

[0053] In summary, the scheme of this embodiment has fast, non-destructive and stable banana quality grading and sorting detection performance, and performs intelligent grading and sorting for banana post-harvest quality, thereby improving its market competitiveness and product added value. At the same time, a large amount of labor is saved, economic costs are saved, and the accuracy, stability and specificity of banana post-harvest quality grading and sorting are better than manual grading and sorting. With the increasing number of samples in the model database, the model algorithm continuously trains, optimizes and verifies the training set, completes the accuracy, stability and specificity of the model in predicting grading and sorting of banana samples of different varieties, and improves the versatility of the model. Moreover, the model's ability to predict the comprehensive quality of samples is far superior to the previous detection method for a single sample and a single quality, and the test results are more scientific, effective and comprehensive.

[0054] Embodiment 2

[0055] This embodiment proposes a nondestructive testing system for comprehensive quality of post-harvest bananas, which is applied to bananas after cold storage, and includes:

[0056] A hyperspectral camera is used to collect hyperspectral images of several bananas and extract spectral features of preset points in the hyperspectral images;

[0057] An infrared thermal imaging camera is used to collect infrared radiation energy distribution images of a number of bananas and convert the infrared radiation energy distribution images into visible light images;

[0058] An intelligent electrical parameter detection unit, used to obtain dielectric characteristic parameters of a number of bananas;

[0059] The data fusion unit is used to fuse the hyperspectral image, spectral features, infrared radiation energy distribution image, visible light image and dielectric characteristic parameters at the data level, define them as sample data, associate the sample data with mature indicators, build a prediction model, select feature variables for fusion, and define them as sample feature variables;

[0060] The model training unit is used to perform decision-level fusion of sample feature variables and maturity indicators based on the prediction model, and finally select batches of samples from the sample feature variables for model training, optimization and deep analysis to complete the training task of the prediction model;

[0061] The maturity grading unit is used to re-perform the data collection task for a single target banana, form real-time sample data and input it into the prediction model to obtain the corresponding real-time maturity index, and use the real-time maturity index as the grading condition for the maturity of the banana.

[0062] The components in this embodiment can refer to S1-S6 in the first embodiment, and will not be described again here.

[0063] Embodiment 3

[0064] This embodiment proposes a nondestructive testing system for the comprehensive quality of post-harvest bananas, which is applied to bananas after cold storage, such as Figure 2 As shown, it includes: a first test dark box 13, a second test dark box 20, and a conveyor belt 7 that passes through the first test dark box 13 and the second test dark box 20 in sequence, an intelligent electrical parameter detection unit 9 is also provided at the end of the conveyor belt 7, a lifting platform 1 is provided between the first test dark box 13 and the second test dark box 20, a hyperspectral camera 2 is provided on one side of the lifting platform 1, and the hyperspectral camera 2 is connected to the first test dark box 13 through an optical fiber, wherein the first test dark box 13 has a built-in halogen light source 15, and the second test dark box 20 has a built-in annular light source 18, an infrared thermal imaging camera 19, a strip light source 21 and an electrical parameter measurement platform 22, and the data output ends of the hyperspectral camera 2, the infrared thermal imaging camera 19 and the electrical parameter measurement platform 22 are connected to the data receiving end of the computer 3.

[0065] Furthermore, a weighing platform 16 is provided at the bottom of the first test dark box 13 , and a weighing device 8 is also provided at the end of the conveyor belt 7 .

[0066] Usage method: Turn on the power supply 5, connect the hyperspectral camera 2 and the conveyor belt 7 to the computer 3 via the USB cable 4, and drive the conveyor belt 7 with the motor 6. The conveyor belt 7 transports the banana sample 17 to the first test dark box 13. The first test dark box 13 has a built-in weighing platform 16. The weighing platform 16 has a built-in weighing sensor under the conveyor belt 7. The weighing sensor is connected to the weighing device 8 externally, and the quality information of the banana sample 17 can be transmitted to the computer 3. The intelligent electrical parameter detection unit 9 is connected to the conveyor belt 7 externally, and is used in conjunction with the electrical parameter measurement platform 22 in the second test dark box 20.

[0067] In a specific embodiment, Figure 3 As shown, the first test dark box 13 has a cooling fan 14, a halogen light source 15, and a weighing platform 16 built in. The halogen light source 15 is arranged inside the first test dark box 13 according to 45° illumination, and the cooling fan 14 is placed on both sides of the bottom of the first test dark box 13. The test bench 11 is threadedly connected to the lifting platform 1, and the lifting platform 1 is threadedly connected to the hyperspectral camera 2. According to the test requirements, the lifting platform 1 is used to control the height of the hyperspectral camera 2, and the hyperspectral camera 2 is used in conjunction with the hyperspectral camera lens 10 to adjust the focal length and other parameters. The optical fiber 12 is connected to the inside of the first test dark box 13 to obtain the hyperspectral image of the banana sample 17.

[0068] In a specific embodiment, Figure 4As shown, the second test dark box 20 has a built-in annular light source 18, an infrared thermal imaging camera 19, a strip light source 21 and an electrical parameter measurement platform 22. The infrared thermal imaging camera 19 is connected to the computer 3 using a USB cable to obtain information such as the infrared radiation energy distribution image and the visible light image of the banana sample 17. The electrical parameter measurement platform 22 cooperates with the intelligent electrical parameter detection unit 9 to obtain the dielectric characteristic parameters of the banana sample 17 and transmits them to the computer 3.

[0069] FigSpec spectrum acquisition software is installed on computer 3 to collect the spectrum information of banana sample 17 in the 400-1000nm band, i.e., the hyperspectral image, and the 400-1000nm spectrum data of any point in the image. The region of interest of the hyperspectral image is selected, and the average spectrum is extracted as the sample spectrum feature information through spectral preprocessing, such as Figure 5 shown.

[0070] For more specific detection steps, please refer to Example 1.

[0071] In addition to being able to accurately classify bananas after cold storage, this solution can also reduce the workload of operators, reduce labor costs, and achieve batch post-harvest quality testing of bananas, thereby improving the market competitiveness of bananas. In addition, the testing method of this solution is universal and can be used for non-destructive testing of the comprehensive quality of various types of bananas on the market. It is effective not only for fresh bananas, but also for cooked bananas.

[0072] The above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable ordinary technicians in the field to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any equivalent changes or modifications made based on the essence of the content of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for nondestructive testing of the comprehensive quality of post-harvest bananas, applied to bananas after cold storage, characterized in that: The following steps are involved: S1, collecting hyperspectral images of a number of bananas, and extracting spectral features of preset points in the hyperspectral images; S2, collecting infrared radiation energy distribution images of a number of bananas, and converting the infrared radiation energy distribution images into visible light images; S3, obtaining dielectric property parameters of a number of bananas; S4, fusing the hyperspectral image, the spectral feature, the infrared radiation energy distribution image, the visible light image and the dielectric characteristic parameter at the data level to define the sample data, associating the sample data with the maturity index, building a prediction model, selecting feature variables for fusion, and defining the sample feature variables; S5, based on the prediction model, performing decision-level fusion on the sample feature variables and the maturity index, and finally selecting batches of samples from the sample feature variables for model training, optimization and deep analysis, to complete the training task of the prediction model; S6, re-performing the data collection tasks of S1-S3 for a single target banana, forming real-time sample data and inputting the data into the prediction model to obtain a corresponding real-time maturity index, and using the real-time maturity index as a grading condition for the maturity of the banana.

2. The method for nondestructive testing of the comprehensive quality of postharvest bananas as claimed in claim 1, characterized in that: A region of interest of the hyperspectral image is selected, and spectral preprocessing is performed on the region of interest, and an average spectrum is extracted as the spectral feature.

3. The method for nondestructive testing of the comprehensive quality of postharvest bananas as claimed in claim 1, characterized in that: The maturity index at least includes soluble solid content, titratable acid, sugar-acid ratio and / or hardness.

4. The method for nondestructive testing of comprehensive quality of postharvest bananas as claimed in claim 1, characterized in that: The infrared radiation energy distribution image is converted into a visible light image through spectrum conversion and brightness enhancement, and pixel values ​​of RGB three primary color channels of the visible light image are extracted. Then, the hue, saturation and brightness pixel values ​​of the visible light image are extracted using a color model, and the sample texture of the visible light image is extracted to estimate the surface area of ​​the banana.

5. The method for nondestructive testing of comprehensive quality of postharvest bananas as claimed in claim 1, characterized in that: The dielectric characteristic parameters include complex impedance and complex dielectric constant, wherein the complex impedance is a physical constant of banana, and the complex dielectric constant is used to characterize the internal quality characteristics of banana.

6. The method for nondestructive testing of comprehensive quality of postharvest bananas as claimed in claim 1, characterized in that: The method further includes S7, presetting the banana maturity level, performing semi-supervised deep learning and representation learning on the banana maturity level through a prediction model, performing translation-invariant classification on the input real-time sample data according to the hierarchical structure of the banana maturity level, and then transmitting the target bananas to a designated channel for packing through a sorting mechanism.

7. A nondestructive testing system for the comprehensive quality of post-harvest bananas, applied to bananas after cold storage, characterized in that: include: A hyperspectral camera is used to collect hyperspectral images of a plurality of bananas and extract spectral features of preset points in the hyperspectral images; An infrared thermal imaging camera, used for collecting infrared radiation energy distribution images of a plurality of bananas, and converting the infrared radiation energy distribution images into visible light images; An intelligent electrical parameter detection unit, used to obtain dielectric characteristic parameters of a number of bananas; A data fusion unit, used for fusing the hyperspectral image, the spectral feature, the infrared radiation energy distribution image, the visible light image and the dielectric characteristic parameter at the data level, defining the result as sample data, associating the sample data with the maturity index, building a prediction model, selecting feature variables for fusion, and defining the result as sample feature variables; A model training unit is used to perform decision-level fusion of the sample feature variables and the maturity index based on the prediction model, and finally select batches of samples from the sample feature variables for model training, optimization and deep analysis to complete the training task of the prediction model; The maturity grading unit is used to re-perform the data collection task for a single target banana, form real-time sample data and input it into the prediction model to obtain a corresponding real-time maturity index, and use the real-time maturity index as a grading condition for the maturity of the banana.

8. A non-destructive testing system for the comprehensive quality of post-harvest bananas, applied to bananas after cold storage, characterized in that: include: A first test dark box, a second test dark box, and a conveyor belt passing through the first test dark box and the second test dark box in sequence, an intelligent electrical parameter detection unit is also arranged at the end of the conveyor belt, a lifting platform is arranged between the first test dark box and the second test dark box, a hyperspectral camera is arranged on one side of the lifting platform, and the hyperspectral camera is connected to the first test dark box through an optical fiber, wherein the first test dark box has a built-in halogen light source, and the second test dark box has a built-in annular light source, an infrared thermal imaging camera, a strip light source and an electrical parameter measurement platform, and the data output ends of the hyperspectral camera, the infrared thermal imaging camera and the electrical parameter measurement platform are connected to the data receiving end of a computer.

9. The method for nondestructive testing of comprehensive quality of post-harvest bananas as claimed in claim 8, characterized in that: A weighing platform is arranged at the bottom of the first test dark box.

10. The method for nondestructive testing of comprehensive quality of post-harvest bananas according to claim 8, characterized in that: A weighing device is also provided at the end of the conveyor belt.

Images