CN115950833A - Method, system, equipment and medium for detecting quality of processed tomatoes - Google Patents
Method, system, equipment and medium for detecting quality of processed tomatoes Download PDFInfo
- Publication number
- CN115950833A CN115950833A CN202310034407.5A CN202310034407A CN115950833A CN 115950833 A CN115950833 A CN 115950833A CN 202310034407 A CN202310034407 A CN 202310034407A CN 115950833 A CN115950833 A CN 115950833A
- Authority
- CN
- China
- Prior art keywords
- detection
- processed
- quality
- neural network
- sample data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture
Landscapes
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention discloses a method, a system, equipment and a medium for detecting the quality of processed tomatoes, which relate to the field of quality detection of the processed tomatoes, and the method comprises the following steps: acquiring a hyperspectral image of a processed tomato to be detected; calculating the average spectral reflectivity according to the hyperspectral image; inputting the average spectral reflectivity into detection models with different qualities for processing tomatoes to obtain detection results; the detection result comprises the soluble solid content, lycopene content, hardness, titratable acid content and maturity of the processed tomato to be detected; the different quality detection models are obtained by training different recurrent neural network models by using different sample data; the sample data comprises the average spectral reflectance and corresponding quality labels for a plurality of processed tomato samples. By adopting the method for detecting the quality of the processed tomatoes, the quality of the processed tomatoes can be detected at the same time, the diversity of monitoring indexes is improved, and the detection efficiency is improved.
Description
Technical Field
The invention relates to the field of quality detection of processed tomatoes, in particular to a method, a system, equipment and a medium for detecting the quality of the processed tomatoes.
Background
The processed tomato fruits are oval in shape, thick in peel, and resistant to storage and transportation, contain rich antioxidants such as lycopene and carotenoid, have high nutritional value, and are main raw materials for preparing tomato paste, preserved fruits and lycopene. The quality of the fruits is closely related to the harvesting time, the quality of the fruits can be reduced when the fruits are harvested too early, the storage time of the fruits can be reduced when the fruits are harvested too late, and resource waste is caused, so that the detection of the maturity is of great significance for ensuring the quality of the fruits. Hardness, soluble solids, titratable acid, lycopene content are not only four important indicators in determining the quality of the processed tomato fruit, but also are closely related to the ripeness of the fruit. The quality of tomato products can be improved by selecting excellent processed tomato raw materials, and the traditional quality detection method has the advantages of single detection index, time and labor consumption, strong destructiveness and difficulty in batch detection of fruits.
The hyperspectral imaging technology is based on multiband image data technology, can simultaneously obtain spatial image information and spectral information of a measured object, and can perform rapid and nondestructive analysis on a research object. Deep learning is a field of machine learning, aims to learn the intrinsic rules and expression levels of sample data, and can extract the features of the data and learn the rules in the data. The cyclic neural network is a deep learning method, is good at mining sequence data and has certain memory capacity. With the help of the storage unit, the output sequence at the current moment can establish a mathematical relationship with all previous sequences, so the recurrent neural network is very effective for data with sequence characteristics, and can mine time sequence and semantic information in the data.
At present, technologies for predicting the quality of agricultural products by using spectral imaging and deep learning technologies exist, but most of previous researches focus on detecting the hardness and solid content of processed tomatoes, and few researches on detecting the maturity, lycopene and titratable acid content of the processed tomatoes are carried out. In addition, when the quality of agricultural products is detected, a classical machine learning model is mostly selected, and a deep learning model is less applied.
Disclosure of Invention
The invention aims to provide a method, a system, equipment and a medium for detecting the quality of processed tomatoes, so as to realize multi-quality detection of the processed tomatoes and improve the diversity of detection indexes.
In order to achieve the purpose, the invention provides the following scheme:
a quality detection method for processed tomatoes comprises the following steps:
acquiring a hyperspectral image of a processed tomato to be detected;
calculating the average spectral reflectivity according to the hyperspectral image;
inputting the average spectral reflectivity into detection models with different qualities of processed tomatoes to obtain detection results; the different quality detection models comprise a soluble solid detection model, a lycopene detection model, a hardness detection model, a titratable acid detection model and a maturity detection model; the detection result comprises the soluble solid content, lycopene content, hardness, titratable acid content and maturity of the processed tomato to be detected;
the different quality detection models are obtained by training a recurrent neural network model by using different sample data; the sample data comprises average spectral reflectance and corresponding quality labels for a plurality of processed tomato samples; the quality label is a soluble solid content label, a lycopene content label, a hardness label, a titratable acid content label or a maturity label; the recurrent neural network model comprises a first recurrent neural network model, a second recurrent neural network model, a third recurrent neural network model, a fourth recurrent neural network model and a fifth recurrent neural network model.
Optionally, the acquiring a hyperspectral image of a processed tomato to be detected specifically includes:
and acquiring two hyperspectral images of the processed tomato to be detected in the opposite direction.
Optionally, the calculating an average spectral reflectance according to the hyperspectral image specifically includes:
calibrating the hyperspectral image to obtain a calibrated hyperspectral image;
extracting an interested region of the calibrated hyperspectral image to obtain an interested region image;
calculating an average spectrum of the region of interest image;
an average spectral reflectance is determined from the average spectrum.
Optionally, the calibrating the hyperspectral image to obtain a calibrated hyperspectral image specifically includes:
using a formula
Calibrating the hyperspectral image to obtain a calibrated hyperspectral image; wherein, C r Representing the calibrated hyperspectral image, C I Representing a hyperspectral image, C b Representing a gray reference image.
Optionally, the training of the recurrent neural network model by using different sample data specifically includes:
training a first circulating neural network model by using first sample data to obtain the soluble solid detection model; the first sample data comprises average spectral reflectivities and corresponding soluble solids content labels for a plurality of processed tomato samples;
training a second recurrent neural network model by using second sample data to obtain the lycopene detection model; the second sample data comprises average spectral reflectance and corresponding lycopene content labels for a plurality of processed tomato samples;
training a third circulating neural network model by using third sample data to obtain the hardness detection model; said third sample data comprises the average spectral reflectance and corresponding hardness labels of a plurality of processed tomato samples;
training a fourth circulating neural network model by using fourth sample data to obtain the titratable acid detection model; said fourth sample data comprises the average spectral reflectance and corresponding titratable acid content labels for a plurality of processed tomato samples;
training a fifth circulating neural network model by using fifth sample data to obtain the maturity detection model; said fifth sample data comprises the average spectral reflectance and corresponding maturity labels of a plurality of processed tomato samples.
A processed tomato quality detection system, comprising:
the image acquisition module is used for acquiring a hyperspectral image of the processed tomato to be detected;
the reflectivity calculation module is used for calculating the average spectral reflectivity according to the hyperspectral image;
the quality detection module is used for inputting the average spectral reflectivity into different quality detection models for processing tomatoes to obtain detection results; the different quality detection models comprise a soluble solid detection model, a lycopene detection model, a hardness detection model, a titratable acid detection model and a maturity detection model; the detection result comprises the soluble solid content, lycopene content, hardness, titratable acid content and maturity of the processed tomato to be detected;
the different quality detection models are obtained by training a recurrent neural network model by using different sample data; the sample data comprises average spectral reflectance and corresponding quality labels for a plurality of processed tomato samples; the quality label is a soluble solid content label, a lycopene content label, a hardness label, a titratable acid content label or a maturity label.
An electronic device, comprising: the device comprises a memory and a processor, wherein the memory is used for storing a computer program, and the processor runs the computer program to enable the electronic equipment to execute the processed tomato quality detection method.
A computer readable storage medium, storing a computer program which, when executed by a processor, implements the above-described method for detecting the quality of processed tomatoes.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
the method for detecting the quality of the processed tomatoes comprises the steps of acquiring hyperspectral images of the processed tomatoes to be detected; calculating the average spectral reflectivity according to the hyperspectral image; inputting the average spectral reflectivity into detection models with different qualities of processed tomatoes to obtain detection results; the detection result comprises the soluble solid content, lycopene content, hardness, titratable acid content and maturity of the processed tomato to be detected; the different quality detection models are obtained by training different recurrent neural network models by using different sample data; the sample data comprises the average spectral reflectance and corresponding quality labels for a plurality of processed tomato samples. The quality detection method for the processed tomatoes can detect multiple qualities of the processed tomatoes at the same time, improves the diversity of detection indexes, and improves the detection efficiency.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a flow chart of a method for detecting the quality of processed tomatoes according to the present invention;
FIG. 2 is a flow chart of a recurrent neural network model training process provided by the present invention;
FIG. 3 is a diagram of a recurrent neural network model architecture provided by the present invention;
fig. 4 is a block diagram of a system for detecting the quality of processed tomatoes provided by the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide a method, a system, equipment and a medium for detecting the quality of processed tomatoes, so as to realize multi-quality detection of the processed tomatoes and improve the diversity of detection indexes.
Based on the method and the system, the invention provides a method and a system for multi-quality prediction and maturity classification of processed tomatoes (namely a method and a system for detecting the quality of the processed tomatoes) based on near-infrared hyperspectral imaging and deep learning, so that the internal relation among different wavelengths is excavated by fully utilizing the memory capacity of a storage unit of a recurrent neural network, the accurate prediction of the multi-quality and maturity of the processed tomatoes is realized, and the detection efficiency is improved.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Example one
As shown in fig. 1, the method for detecting the quality of processed tomatoes comprises the following steps:
step 101: and acquiring a hyperspectral image of the processed tomato to be detected. In practical application, two hyperspectral images of the tomato to be processed in the opposite direction are obtained. The built hyperspectral acquisition system is used for acquiring and processing hyperspectral images of the front surface and the back surface of the tomato, so that the problem of unavailable spectral information caused by overexposure of the surface of the tomato can be reduced.
Step 102: and calculating the average spectral reflectivity according to the hyperspectral image.
Further, the step 102 specifically includes:
and calibrating the hyperspectral image to obtain a calibrated hyperspectral image.
And extracting the interested region of the calibrated hyperspectral image to obtain an interested region image.
Calculating an average spectrum of the region of interest image.
An average spectral reflectance is determined from the average spectrum.
Further, the calibrating the hyperspectral image to obtain a calibrated hyperspectral image specifically includes:
using formulas
Calibrating the hyperspectral image to obtain a calibrated hyperspectral image; wherein, C r Representing the calibrated hyperspectral image, C I Representing a hyperspectral image, C b Representing a gray reference image.
Step 103: and inputting the average spectral reflectivity into detection models with different qualities for processing the tomatoes to obtain detection results. The different quality detection models comprise a soluble solid detection model, a lycopene detection model, a hardness detection model, a titratable acid detection model and a maturity detection model; the detection result comprises the soluble solid content, lycopene content, hardness, titratable acid content and maturity of the processed tomato to be detected.
The different quality detection models are obtained by training a recurrent neural network model by using different sample data; the sample data comprises average spectral reflectance and corresponding quality labels for a plurality of processed tomato samples; the quality label is a soluble solid content label, a lycopene content label, a hardness label, a titratable acid content label or a maturity label; the recurrent neural network model comprises a first recurrent neural network model, a second recurrent neural network model, a third recurrent neural network model, a fourth recurrent neural network model and a fifth recurrent neural network model.
Further, the training of the recurrent neural network model by using different sample data specifically includes:
training a first cyclic neural network model by using first sample data to obtain the soluble solid detection model; the first sample data comprises the average spectral reflectance and corresponding soluble solids content signatures for a plurality of processed tomato samples.
Training a second recurrent neural network model by using second sample data to obtain the lycopene detection model; the second sample data comprises average spectral reflectance and corresponding lycopene content labels for a plurality of processed tomato samples.
Training a third circulating neural network model by using third sample data to obtain the hardness detection model; the third sample data comprises average spectral reflectance and corresponding hardness labels for a plurality of processed tomato samples.
Training a fourth circulating neural network model by using fourth sample data to obtain the titratable acid detection model; the fourth sample data comprises the average spectral reflectance and corresponding titratable acid content labels for a plurality of processed tomato samples.
Training a fifth circulating neural network model by using fifth sample data to obtain the maturity detection model; the fifth sample data comprises the average spectral reflectance and the corresponding maturity label of a plurality of processed tomato samples.
In practical application, the present invention trains different recurrent neural network models, and a specific flow chart is shown in fig. 2, and the training process for different recurrent neural network models is specifically as follows:
step 1: picking tomato samples processed in green ripe period, color conversion period and red ripe period, wiping the picked samples one by one, placing the cleaned tomato samples at an ambient temperature for 12 hours, and performing subsequent spectral image acquisition and physical and chemical tests when the temperature of the processed tomato samples is the same as the room temperature.
And 2, step: the built hyperspectral acquisition system is used for acquiring and processing hyperspectral images of the front surface and the back surface of the tomato, so that the problem of unavailable spectral information caused by overexposure of the surface of the tomato can be reduced. Before the experiment, open high spectrum collection system and preheat, treat high spectrum collection system environmental stability back, will process the tomato sample and put and carry out spectral image collection and calibration on putting the thing platform, calibration formula is as follows:
wherein C is r Representing a reflectance image (calibrated hyperspectral image), C I Representing the original image (acquired hyperspectral image), C b Is a gray reference image comprising 50% black and 50% white.
And step 3: extracting an image of interest: and (3) selecting the hyperspectral images calibrated in the step (2), and eliminating the noise wavelength images before and after the hyperspectral images, wherein the effective wavelength range of the new hyperspectral images is 956nm-1679nm. Selecting an image calculation mask with the wavelength of 1300nm, sequentially extracting single processed tomato samples from the hyperspectral images after the noise wavelengths before and after the elimination of the mask region, and taking the extracted processed tomatoes as images of interest.
And 4, step 4: average spectral acquisition of processed tomatoes: and (4) calculating the average spectrum of each interested region image according to the divided hyperspectral images (interested region images) of the interested regions in the step (3). The average spectrum of the g band is calculated as follows:
A C ＝mean(∑F h，i，g for all(h，i)∈T) (2)
wherein T is the leaf region in the interested region image in the step 3, h and i are the coordinate values of pixel points in the image respectively, h is the abscissa value of the pixel point, i is the ordinate value of the pixel point, g represents the g-th wave band, F h，i，g Is the value of the coordinates (h, i) of the image at the g-th band.
And 5: the saccharimeter (pal-1) and the hardness meter are used for measuring the saccharinity value (soluble solid content) and the hardness of the processed tomatoes, and the acid-base titration method and the extraction colorimetric method are used for measuring the titratable acid content and the lycopene content of the processed tomatoes.
Step 6: and (4) averaging the spectra in step 4 according to a ratio of 4: the ratio of 1 is divided into a correction set and a prediction set for training a recurrent neural network model, as shown in fig. 3, the network structure of the recurrent neural network model comprises a plurality of recurrent units, each recurrent unit comprises an input layer, a hidden layer and an output layer, all the hidden layers are connected, by means of a memory storage unit, by giving different weights between layers and between recurrent bodies, the output sequence of the RNN at the current moment can establish a mathematical relationship with all previous sequences, that is, the output of the hidden layer at the t moment is not only related to the input of the t moment, but also related to the output of the hidden layer at the t-1 moment. The average spectral reflectance result of the processed tomato sample and the reference value (sugar degree value, hardness, titratable acid content or lycopene content) of the experimental measurement are input into the input layer for training, and the output layer is the prediction result of the soluble solid content, hardness, lycopene content or titratable acid content of the processed tomato. The activation function of the hidden layer is tanh, the activation function of the output layer is softmax, and the relation between the hidden layer and the output layer at the moment t is as follows:
yt＝softmax(h t V+b y ) (3)
h t ＝tan h(x t U+h t-1 W+b h ) (4)
where tanh and softmax are activation functions for the hidden layer and the output layer, respectively. U, W and V are weights of the loop node and the output node, the weights are not changed in the loop, bh and b y Is the deviation of the set of hidden and output layers.
And 7: the root mean square error loss function is used in the training of the regression model (different quality detection model), and the cross entropy loss function is used in the classification model (maturity detection model), so that the faster convergence of the neural network is realized.
Where equation (5) is the root mean square error loss function,
is the ith trainingThe label value of the sample, is->
Is the ith training sample. Equation (6) is a cross entropy loss function, based on>
Is the label value of the ith training sample, based on the number of training samples in the past>
Is the ith training sample.
And step 8: training a circulating neural network prediction model by using processed tomato sample data, and inputting the average spectral reflectivity of the processed tomatoes of the sample to be detected into the trained circulating neural network prediction model to obtain prediction results of a soluble solid value, a hardness value, a titratable acid value, a lycopene value and a maturity.
For example, for a certain processed tomato sample, the predicted result of the recurrent neural network model is: SSC:5.32 percent; hardness: 7.34kg/cm -2 (ii) a Lycopene: 8.93mg/100g; titratable acid: 0.33 percent; maturity: red ripe period. The practical result is: SSC:5.40 percent; hardness: 7.53kg/cm -2 (ii) a Lycopene: 8.85mg/100g; titratable acid: 0.32 of; maturity: red ripe period.
And step 9: using decision coefficients (R) of the training set C 2 ) And Root Mean Square Error (RMSEC), evaluating the performance of the quality prediction model, and evaluating the performance of the classification model by using accuracy accuracycacy, precision and recall. The calculation method is as follows:
wherein, y i And
correcting and predicting the true and predicted values, y, of the measured indicators in the set m The average value of the measured indexes in the data set is shown, TP is the real sample number, FP is the false sample number, TN is the real negative sample number, and FN is the false negative sample number.
The invention provides a processed tomato quality detection method based on a hyperspectral imaging technology and deep learning, which utilizes a multi-level structural recurrent neural network to process spectral data and measured quality index reference values, deeply excavates spectral information by giving different weights between layers and between recurrent bodies, and introduces a root mean square error loss function, realizes faster convergence of the neural network, and can learn internal relations among different wavelengths and reference values of the wavelengths and the quality through the memory capacity of a storage unit of the recurrent neural network. The method can quickly and efficiently predict the quality of the processed tomatoes.
The invention provides a method for detecting tomatoes processed in different maturity stages based on a hyperspectral imaging technology and deep learning. The method can rapidly detect the maturity of the processed tomatoes.
The invention provides a method capable of solving the overexposure phenomenon generated in the image acquisition process of smooth-surface fruits.
Based on the technology, the method obtains good results on the prediction of the quality and the maturity of the processed tomatoes. Experiments prove that the correlation coefficient of a training set of the recurrent neural network model is 0.905 and the root mean square error is 0.15 in SSC prediction. On the aspect of hardness prediction, the correlation coefficient of a training set is 0.961, and the root mean square error is 0.66; on titratable acid prediction, the correlation coefficient of a training set is 0.903, and the root mean square error is 0.03; on lycopene prediction, the correlation coefficient of the training set is 0.914, and the root mean square error is 0.18; on the maturity classification, the accuracy rate is 0.966, the precision rate is 0.972, and the recall rate is 0.952.
Example two
In order to implement the corresponding method of the above embodiment to achieve the corresponding functions and technical effects, the following provides a processed tomato quality detection system, as shown in fig. 4, comprising:
the image acquisition module 401 is configured to acquire a hyperspectral image of a processed tomato to be detected.
A reflectivity calculation module 402, configured to calculate an average spectral reflectivity according to the hyperspectral image.
A quality detection module 403, configured to input the average spectral reflectivity into different quality detection models for processing tomatoes to obtain a detection result; the different quality detection models comprise a soluble solid detection model, a lycopene detection model, a hardness detection model, a titratable acid detection model and a maturity detection model; the detection result comprises the soluble solid content, lycopene content, hardness, titratable acid content and maturity of the processed tomato to be detected.
The different quality detection models are obtained by training a recurrent neural network model by using different sample data; the sample data comprises average spectral reflectance and corresponding quality labels for a plurality of processed tomato samples; the quality label is a soluble solid content label, a lycopene content label, a hardness label, a titratable acid content label or a maturity label.
EXAMPLE III
The present invention also provides an electronic device, comprising: the device comprises a memory and a processor, wherein the memory is used for storing a computer program, and the processor runs the computer program to enable the electronic equipment to execute the processed tomato quality detection method in the first embodiment.
Example four
The invention also provides a computer readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the computer program implements the processed tomato quality detection method of the first embodiment.
In the present specification, the embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (8)
1. A method for detecting the quality of processed tomatoes is characterized by comprising the following steps:
acquiring a hyperspectral image of a processed tomato to be detected;
calculating the average spectral reflectivity according to the hyperspectral image;
inputting the average spectral reflectivity into detection models with different qualities for processing tomatoes to obtain detection results; the different quality detection models comprise a soluble solid detection model, a lycopene detection model, a hardness detection model, a titratable acid detection model and a maturity detection model; the detection result comprises the soluble solid content, lycopene content, hardness, titratable acid content and maturity of the processed tomato to be detected;
the different quality detection models are obtained by training a recurrent neural network model by using different sample data; the sample data comprises average spectral reflectance and corresponding quality labels for a plurality of processed tomato samples; the quality label is a soluble solid content label, a lycopene content label, a hardness label, a titratable acid content label or a maturity label; the recurrent neural network model comprises a first recurrent neural network model, a second recurrent neural network model, a third recurrent neural network model, a fourth recurrent neural network model and a fifth recurrent neural network model.
2. The processed tomato quality detection method according to claim 1, wherein the acquiring hyperspectral images of the processed tomatoes to be detected specifically comprises:
and acquiring two hyperspectral images of the processed tomato to be detected in the relative direction.
3. The method for detecting the quality of the processed tomatoes according to claim 1, wherein the calculating the average spectral reflectance according to the hyperspectral image specifically comprises:
calibrating the hyperspectral image to obtain a calibrated hyperspectral image;
extracting an interested region of the calibrated hyperspectral image to obtain an interested region image;
calculating an average spectrum of the region of interest image;
an average spectral reflectance is determined from the average spectrum.
4. The processed tomato quality detection method according to claim 3, wherein the calibrating the hyperspectral image to obtain a calibrated hyperspectral image specifically comprises:
using formulas
Calibrating the hyperspectral image to obtain a calibrated hyperspectral image; wherein, C r Representing the calibrated hyperspectral image, C I Representing a hyperspectral image, C b Representing a gray reference image.
5. The method of claim 1, wherein the training of the recurrent neural network model using different sample data comprises:
training a first circulating neural network model by using first sample data to obtain the soluble solid detection model; the first sample data comprises an average spectral reflectance and a corresponding soluble solids content signature for a plurality of processed tomato samples;
training a second recurrent neural network model by using second sample data to obtain the lycopene detection model; the second sample data comprises average spectral reflectance and corresponding lycopene content labels for a plurality of processed tomato samples;
training a third cyclic neural network model by using third sample data to obtain the hardness detection model; said third sample data comprises the average spectral reflectance and corresponding hardness label of a plurality of processed tomato samples;
training a fourth circulating neural network model by using fourth sample data to obtain the titratable acid detection model; the fourth sample data comprises average spectral reflectance and corresponding titratable acid content labels for a plurality of processed tomato samples;
training a fifth circulating neural network model by using fifth sample data to obtain the maturity detection model; said fifth sample data comprises the average spectral reflectance and corresponding maturity labels of a plurality of processed tomato samples.
6. A processed tomato quality detection system, comprising:
the image acquisition module is used for acquiring a hyperspectral image of the processed tomato to be detected;
the reflectivity calculation module is used for calculating the average spectral reflectivity according to the hyperspectral image;
the quality detection module is used for inputting the average spectral reflectivity into different quality detection models for processing tomatoes to obtain detection results; the different quality detection models comprise a soluble solid detection model, a lycopene detection model, a hardness detection model, a titratable acid detection model and a maturity detection model; the detection result comprises the soluble solid content, lycopene content, hardness, titratable acid content and maturity of the processed tomato to be detected;
the different quality detection models are obtained by training a recurrent neural network model by using different sample data; the sample data comprises average spectral reflectance and corresponding quality labels for a plurality of processed tomato samples; the quality label is a soluble solid content label, a lycopene content label, a hardness label, a titratable acid content label or a maturity label.
7. An electronic device, comprising: a memory for storing a computer program and a processor for executing the computer program to cause the electronic device to perform the method of quality detection of processed tomatoes as claimed in any one of claims 1 to 5.
8. A computer-readable storage medium, wherein a computer program is stored, which when executed by a processor, implements the method for quality testing of processed tomatoes of any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310034407.5A CN115950833A (en) | 2023-01-10 | 2023-01-10 | Method, system, equipment and medium for detecting quality of processed tomatoes |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310034407.5A CN115950833A (en) | 2023-01-10 | 2023-01-10 | Method, system, equipment and medium for detecting quality of processed tomatoes |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115950833A true CN115950833A (en) | 2023-04-11 |
Family
ID=87286053
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310034407.5A Pending CN115950833A (en) | 2023-01-10 | 2023-01-10 | Method, system, equipment and medium for detecting quality of processed tomatoes |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115950833A (en) |
-
2023
- 2023-01-10 CN CN202310034407.5A patent/CN115950833A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Pu et al. | Hierarchical variable selection for predicting chemical constituents in lamb meats using hyperspectral imaging | |
| Sabilla et al. | Estimating gas concentration using artificial neural network for electronic nose | |
| CN110455722A (en) | Rubber tree blade phosphorus content EO-1 hyperion inversion method and system | |
| Srivastava et al. | Data processing approaches and strategies for non-destructive fruits quality inspection and authentication: a review | |
| CN108169165B (en) | Maltose mixture quantitative analysis method based on terahertz spectrum and image information fusion | |
| Jie et al. | Research on Citrus grandis granulation determination based on hyperspectral imaging through deep learning | |
| CN111879709B (en) | Lake water body spectral reflectivity inspection method and device | |
| CN115443889A (en) | Accurate irrigation method and device for crops | |
| Chen et al. | Rapid detection of pomelo fruit quality using near-infrared hyperspectral imaging combined with chemometric methods | |
| Guo et al. | Method and system for nondestructive detection of freshness in Penaeus vannamei based on hyperspectral technology | |
| Wei et al. | Prediction of the maturity of greenhouse grapes based on imaging technology | |
| CN111445469A (en) | Hyperspectrum-based apple damage parameter lossless prediction method after impact | |
| Martins et al. | Estimation of soluble solids content and fruit temperature in'Rocha'pear using Vis-NIR spectroscopy and the SpectraNet–32 deep learning architecture | |
| Ekramirad et al. | Application of Hyperspectral imaging and acoustic emission techniques for apple quality prediction | |
| CN110779875A (en) | Method for detecting moisture content of winter wheat ear based on hyperspectral technology | |
| CN115950833A (en) | Method, system, equipment and medium for detecting quality of processed tomatoes | |
| Chia et al. | Neural network and extreme gradient boosting in near infrared spectroscopy | |
| Zhao et al. | Determination of apple firmness using hyperspectral imaging technique and multivariate calibrations | |
| Zhou et al. | Hyperspectral imaging technology for detection of moisture content of tomato leaves | |
| CN115049902A (en) | Citrus leaf water content visual prediction method, system, equipment and storage medium | |
| Pamornnak et al. | Oil content determination scheme of postharvest oil palm for mobile devices | |
| CN113496218A (en) | Evaluation method and system for hyperspectral remote sensing sensitive band selection mode | |
| Azis et al. | Classification of dried chilli quality using image processing | |
| CN103592258A (en) | Detection method of tea polyphenol contents in tea | |
| Pourdarbani et al. | Non-destructive Estimation of Chemical Properties in Braeburn Apple using Convolutional Neural Network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |