CN114266419A - Cigar tobacco leaf process stage prediction method, system and medium based on data fusion - Google Patents
Cigar tobacco leaf process stage prediction method, system and medium based on data fusion Download PDFInfo
- Publication number
- CN114266419A CN114266419A CN202210032426.XA CN202210032426A CN114266419A CN 114266419 A CN114266419 A CN 114266419A CN 202210032426 A CN202210032426 A CN 202210032426A CN 114266419 A CN114266419 A CN 114266419A
- Authority
- CN
- China
- Prior art keywords
- data
- processing
- original
- cigar tobacco
- image
- 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
Landscapes
- Manufacture Of Tobacco Products (AREA)
Abstract
The invention belongs to the technical field of cigar airing, and discloses a method, a system and a medium for predicting the process stage of cigar tobacco leaves based on data fusion, wherein the prediction method comprises the following steps: acquiring original data and original images in the tobacco leaf airing process of the cigar in a preset acquisition period; respectively processing original data and original images, training a decision tree model and a support vector machine model based on the processed images and data, and fusing prediction results by using a weighted voting method; and acquiring original data and original images of the cigar tobacco leaves in the airing process in real time, processing the images to be used as model input items, and distinguishing the cigar tobacco leaves in the airing stage. The invention provides a cigar tobacco leaf process stage prediction method based on data fusion, which can accurately judge the cigar tobacco leaf airing state, is convenient for humidity management of an airing room, improves the airing process and saves the labor cost.
Description
Technical Field
The invention belongs to the technical field of cigar airing, and particularly relates to a method, a system and a medium for predicting the process stage of cigar tobacco leaves based on data fusion.
Background
At present, the air-curing period is a key period for the quality of tobacco leaves, particularly the appearance quality. At present, the research on the air-curing technology of the cigar tobacco leaves in China is less, a plurality of traditional tobacco areas adopt a sun-curing method (or semi-curing and semi-curing) to cure the cigar tobacco leaves, the curing technical parameters and the curing method mainly stay in the traditional experience and custom summary, or are regarded as an important technical secret without declaration, and mature air-curing technology and matched air-curing facilities are lacked. In the process of airing the cigar tobacco leaves, the tobacco leaf state is judged mainly based on human eyes and contact, the information source is single, and the subjectivity is strong; the tobacco leaf airing state can not be observed constantly in the airing period, and the labor intensity is high.
With the gradual maturity of machine vision and the continuous improvement of sensor performance, can all-round acquisition to the data of cigar system in process of drying in the air with the help of the information acquisition equipment of different functions.
Through the above analysis, the problems and defects of the prior art are as follows: in the prior art, the tobacco leaf airing of the cigar is mainly judged manually, and the method has the advantages of strong subjectivity, large error, high labor intensity, inaccuracy and low intelligent degree.
The difficulty in solving the above problems and defects is:
how to reduce the manual participation in the whole process state judgment of the cigar tobacco leaf airing and reduce the labor intensity; how to accurately acquire and analyze the tobacco leaf airing state of the cigar by means of multiple information sources; how to further mine the collected data and improve the technical bottleneck of the cigar airing in an intelligent mode.
The significance of solving the problems and the defects is as follows: by researching and analyzing the acquired data, the intelligent air-curing of the automatic control of the process and the automatic control of the environmental parameters in the air-curing process of the cigars can be realized, the air-curing labor intensity of the cigars is reduced, the bottleneck of the air-curing technology is broken through, the high-quality development of the cigar industry is promoted by supplementing the cigar technology and the raw material short boards, and the method has very important significance.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a cigar tobacco leaf process stage prediction method, system and medium based on data fusion.
The invention is realized in such a way that a cigar tobacco leaf process stage prediction method based on data fusion comprises the following steps:
acquiring original data and original images in the tobacco leaf airing process of the cigar in a preset acquisition period; respectively processing original data and original images, training a decision tree model and a support vector machine model based on the processed images and data, and fusing prediction results by using a weighted voting method; and acquiring original data and original images of the cigar tobacco leaves in the airing process in real time, processing the images to be used as model input items, and distinguishing the cigar tobacco leaves in the airing stage.
Further, the data fusion-based cigar tobacco leaf process stage prediction method comprises the following steps:
dividing cigars to be aired into different types according to attributes and varieties; when the cigars are aired, acquiring airing room environment data and cigar tobacco leaf data in the airing process in a certain acquisition period to obtain original data and an original image;
step two, performing unstructured data preprocessing on the collected original images of the cigar tobacco leaves to obtain primary processed images and secondary processed images;
step three, cutting the obtained secondary processing image to obtain a tertiary processing image; performing data fusion on the obtained three-time processed image and the original data to obtain primary processed data;
processing the primary processing data to sequentially obtain secondary processing data, tertiary processing data, quartic processing data and quintic processing data;
importing the five-time processing data into a decision tree model and a support vector machine model, and fusing a prediction result by using a weighted voting method according to the performance of the models;
and step six, acquiring original data and original images of the cigar tobacco leaves in the airing process in real time, processing the original data and the original images to be used as input items of a decision tree model and a support vector machine model, and distinguishing the cigar tobacco leaves in the airing stage.
Further, the environmental data includes air-drying humidity; the cigar tobacco leaf data comprises tobacco leaf air-curing images and water loss data.
Further, the acquiring of the room-airing environment data and the cigar tobacco data in the airing process in a certain acquisition period to obtain the original data and the original image comprises:
collecting an air-cured image of the cigar tobacco leaves as an original image by utilizing a camera in a preset collection period; and collecting the air-curing water loss amount of the cigar tobacco leaves and the humidity data of the air-curing room as original data by utilizing each sensor in a preset collection period.
Further, the acquiring of the house-airing environment data and the cigar tobacco data in the airing process in a certain acquisition period to obtain the original data and the original image further comprises:
unifying the format of the collected original data; unifying the format of the collected original images.
Further, the step of preprocessing unstructured data of the acquired original images of the cigar tobacco leaves to obtain primary processed images and secondary processed images comprises the following steps:
determining pixels with similar neighborhood structures in an original image by using a non-local mean image noise reduction algorithm and taking the pixels as units; then, the determined area is weighted and averaged to obtain a primary processing image;
and performing color matching correction on the primary processed image by using an image color correction algorithm and by using a color histogram of a standard color correction card to obtain a secondary processed image.
Further, the obtained secondary processing image is cut to obtain a tertiary processing image; performing data fusion on the obtained three processed images and the original data to obtain primary processed data, wherein the primary processed data comprises the following steps:
cutting a certain fixed area in the center of the obtained secondary processing image to obtain a tertiary processing image;
calculating L, a and b color values of each pixel in the three processed images by using an OpenCV calculation visual library, and taking the average value as an L, a and b color average value of the three processed images; wherein L represents lightness, a represents a component from green to red, b represents a component from blue to yellow;
and fusing the characteristics of L, a and b extracted from the three processed images with the original data by taking time as a corresponding relation to obtain primary processed data.
Further, the processing the primary processing data to sequentially obtain secondary processing data, tertiary processing data, quartic processing data and quintic processing data includes:
according to the corresponding relation between the air-curing time and the air-curing stage in the air-curing process of the cigar tobacco leaves, marking each primary processing data with a corresponding air-curing stage, and dividing the primary processing data into a withering stage, a yellowing stage, a browning stage and a tendon drying stage to obtain secondary processing data;
when a small amount of missing characteristic values exist in the secondary processing data, performing interpolation filling by using the mean values of the adjacent characteristic values before and after the small amount of missing characteristic values to obtain tertiary processing data;
abnormal value identification is carried out on the three-time processed data by using a standard deviation detection method, and the abnormal value is corrected by using the functional relation of the data before and after the abnormal value to obtain four-time processed data;
the data of the four treatments were normalized by the Z-Score normalization method to obtain five treatments.
Further, the importing the five-time processed data into a decision tree model and a support vector machine model, and fusing prediction results by using a weighted voting method according to the performance of the models comprises:
importing the five processed data into a decision tree model for training and prediction comprises the following steps:
taking the water loss amount of tobacco leaf in air-curing, the humidity of an air-curing room, L, a and b as input characteristics, and taking the air-curing stage as an input label; determining input characteristics by using the information gain ratio in the training process, and constructing a decision tree; taking the obtained classification probability and the classification result as a first result, and marking as R1;
importing the five processed data into a support vector machine model for training and prediction comprises the following steps:
taking the water loss amount of tobacco leaf in air-curing, the humidity of an air-curing room, L, a and b as input characteristics, and taking the air-curing stage as an input label; the training process uses a grid searching method to automatically adjust model parameters; taking the obtained classification probability and the classification result as a second result, and marking as R2;
and carrying out weight division on the models according to the accuracy ranking, testing the prediction accuracy after multi-model fusion under different weights, and selecting the most appropriate weight proportion.
Another object of the present invention is to provide a system for predicting the process stage of cigar tobacco based on data fusion, which comprises:
the system comprises an original data and original image acquisition module, a data processing module and a data processing module, wherein the original data and original image acquisition module is used for dividing cigars to be aired into different types according to attributes and varieties; when the cigars are aired, acquiring airing room environment data and cigar tobacco leaf data in the airing process in a certain acquisition period to obtain original data and an original image;
the processing image acquisition module is used for carrying out unstructured data preprocessing on the acquired original images of the cigar tobacco leaves to obtain primary processing images and secondary processing images;
the line data fusion module is used for cutting the obtained secondary processing image to obtain a tertiary processing image; performing data fusion on the obtained three-time processed image and the original data to obtain primary processed data;
the multi-processing data acquisition module is used for processing the primary processing data to sequentially obtain secondary processing data, tertiary processing data, quartic processing data and quintic processing data;
the fusion prediction result module is used for importing the five-time processing data into a decision tree model and a support vector machine model and fusing prediction results by using a weighted voting method according to the performance of the models;
and the judging module is used for acquiring original data and original images of the cigar tobacco leaves in the airing process in real time, processing the original images to be used as input items of a decision tree model and a support vector machine model, and judging the cigar tobacco leaves in the airing stage.
It is a further object of the present invention to provide a computer readable storage medium storing a computer program which, when executed by a processor, causes the processor to execute the data fusion-based cigar tobacco process stage prediction method.
By combining all the technical schemes, the invention has the advantages and positive effects that:
the invention provides a cigar tobacco leaf process stage prediction method based on data fusion, which can accurately judge the cigar tobacco leaf airing state through the output result of a fusion prediction model, avoids the problems of subjectivity, error and high labor intensity in the manual judgment of the cigar tobacco leaf airing stage, facilitates the humidity management of an airing house and improves the airing process.
According to the method, L, a and b color values are extracted from the cigar tobacco leaf airing image and converted into structured data, the structured data and the original data are subjected to data fusion by taking time as a corresponding relation, and the training efficiency of a prediction model is improved;
the method is based on data fusion, obtains the prediction result of the fusion model by utilizing the weight value combination, can judge the tobacco leaf airing stage state of the cigar in real time and in the whole process, improves the judging accuracy and the judging efficiency, reduces the working intensity of airing personnel and saves the labor cost.
Drawings
Fig. 1 is a schematic diagram of a data fusion-based cigar tobacco process stage prediction method provided by an embodiment of the invention.
Fig. 2 is a flow chart of a method for predicting the tobacco processing stage of a cigar based on data fusion according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Aiming at the problems in the prior art, the invention provides a cigar tobacco leaf process stage prediction method based on data fusion, and the invention is described in detail below with reference to the accompanying drawings.
The method for predicting the process stage of the cigar tobacco based on data fusion provided by the embodiment of the invention comprises the following steps:
acquiring original data and original images in the tobacco leaf airing process of the cigar in a preset acquisition period; respectively processing original data and original images, training a decision tree model and a support vector machine model based on the processed images and data, and fusing prediction results by using a weighted voting method; and acquiring original data and original images of the cigar tobacco leaves in the airing process in real time, processing the images to be used as model input items, and distinguishing the cigar tobacco leaves in the airing stage.
As shown in fig. 1, the method for predicting the process stage of the cigar tobacco based on data fusion provided by the embodiment of the invention comprises the following steps:
s101, dividing cigars to be aired into different types according to attributes and varieties; when the cigars are aired, acquiring airing room environment data and cigar tobacco leaf data in the airing process in a certain acquisition period to obtain original data and an original image;
s102, performing unstructured data preprocessing on the collected original images of the cigar tobacco leaves to obtain primary processed images and secondary processed images; cutting the obtained secondary processing image to obtain a tertiary processing image; performing data fusion on the obtained three-time processed image and the original data to obtain primary processed data;
s103, processing the primary processing data to sequentially obtain secondary processing data, tertiary processing data, quartic processing data and quintic processing data; importing the five-time processing data into a decision tree model and a support vector machine model, and fusing a prediction result by using a weighted voting method according to the performance of the models;
and S104, acquiring original data and original images of the cigar tobacco leaves in the airing process in real time, processing the data to be used as input items of a decision tree model and a support vector machine model, and judging the cigar tobacco leaves in the airing stage.
The environmental data provided by the embodiment of the invention comprises airing room humidity; the cigar tobacco leaf data comprises tobacco leaf air-curing images and water loss data.
The method for acquiring the environment data of the air-curing house and the tobacco data of the cigars in the air-curing process in a certain acquisition period to obtain the original data and the original image comprises the following steps:
collecting an air-cured image of the cigar tobacco leaves as an original image by utilizing a camera in a preset collection period; and collecting the air-curing water loss amount of the cigar tobacco leaves and the humidity data of the air-curing room as original data by utilizing each sensor in a preset collection period.
The method for acquiring the environment data of the air-curing house and the tobacco data of the cigars in the air-curing process to obtain the original data and the original image in a certain acquisition period further comprises the following steps:
unifying the format of the collected original data; unifying the format of the collected original images.
The method for preprocessing unstructured data of the acquired cigar tobacco leaf primary image to obtain the primary processing image and the secondary processing image comprises the following steps:
determining pixels with similar neighborhood structures in an original image by using a non-local mean image noise reduction algorithm and taking the pixels as units; then, the determined area is weighted and averaged to obtain a primary processing image;
and performing color matching correction on the primary processed image by using an image color correction algorithm and by using a color histogram of a standard color correction card to obtain a secondary processed image.
The method provided by the embodiment of the invention cuts the obtained secondary processing image to obtain a tertiary processing image; performing data fusion on the obtained three processed images and the original data to obtain primary processed data, wherein the primary processed data comprises the following steps:
cutting a certain fixed area in the center of the obtained secondary processing image to obtain a tertiary processing image;
calculating L, a and b color values of each pixel in the three processed images by using an OpenCV calculation visual library, and taking the average value as an L, a and b color average value of the three processed images; wherein L represents lightness, a represents a component from green to red, b represents a component from blue to yellow;
and fusing the characteristics of L, a and b extracted from the three processed images with the original data by taking time as a corresponding relation to obtain primary processed data.
The embodiment of the invention provides a method for processing primary processing data to sequentially obtain secondary processing data, tertiary processing data, quartic processing data and quintic processing data, which comprises the following steps:
according to the corresponding relation between the air-curing time and the air-curing stage in the air-curing process of the cigar tobacco leaves, marking each primary processing data with a corresponding air-curing stage, and dividing the primary processing data into a withering stage, a yellowing stage, a browning stage and a tendon drying stage to obtain secondary processing data;
when a small amount of missing characteristic values exist in the secondary processing data, performing interpolation filling by using the mean values of the adjacent characteristic values before and after the small amount of missing characteristic values to obtain tertiary processing data;
abnormal value identification is carried out on the three-time processed data by using a standard deviation detection method, and the abnormal value is corrected by using the functional relation of the data before and after the abnormal value to obtain four-time processed data;
the data of the four treatments were normalized by the Z-Score normalization method to obtain five treatments.
The method for importing the five-time processing data into the decision tree model and the support vector machine model and fusing the prediction results by using the weighted voting method according to the performance of the models comprises the following steps:
importing the five processed data into a decision tree model for training and prediction comprises the following steps:
taking the water loss amount of tobacco leaf in air-curing, the humidity of an air-curing room, L, a and b as input characteristics, and taking the air-curing stage as an input label; determining input characteristics by using the information gain ratio in the training process, and constructing a decision tree; taking the obtained classification probability and the classification result as a first result, and marking as R1;
importing the five processed data into a support vector machine model for training and prediction comprises the following steps:
taking the water loss amount of tobacco leaf in air-curing, the humidity of an air-curing room, L, a and b as input characteristics, and taking the air-curing stage as an input label; the training process uses a grid searching method to automatically adjust model parameters; taking the obtained classification probability and the classification result as a second result, and marking as R2;
and carrying out weight division on the models according to the accuracy ranking, testing the prediction accuracy after multi-model fusion under different weights, and selecting the most appropriate weight proportion.
The technical solution of the present invention is further described with reference to the following specific embodiments.
Example 1:
a cigar tobacco leaf process stage prediction method based on data fusion is mainly applied to a cigar tobacco leaf airing process, and the specific implementation scheme is as follows:
s1: before the cigars are aired, different types are divided according to attributes and varieties; in the cigar airing process, collecting airing room environment data and cigar leaf data in the airing process at a certain collecting period to form original data and an original image, wherein the environment data comprises airing room humidity; the cigar leaf data includes leaf curing images and water loss data.
Specifically, in the cigar airing process, the camera collects the airing images of the cigar tobacco leaves in a preset collection period; each sensor collects the air-curing water loss amount of the cigar tobacco leaves and the humidity data of the air-curing room in the same collection period; and then uploading the data to a cloud platform database and a corresponding picture folder through a network to form original data and an original image for model training. The preset collection period is, for example, 30 minutes, 40 minutes, 50 minutes, 60 minutes, or 90 minutes, and the numerical values are only for illustration and are not limited specifically.
S2: and (3) performing unstructured data preprocessing on the acquired original images of the cigar tobacco leaves to obtain secondary processed images.
Specifically, the original image of the cigar tobacco leaves collected in step S1 is obtained, a non-local mean image denoising algorithm is adopted, pixels with similar neighborhood structures are searched in the original image by taking the pixels as units, weighted average is then obtained for the areas, gaussian noise in the original image is removed to a greater extent, and a primary processed image is obtained; and performing color correction by adopting an image color correction algorithm and performing color matching on the primary processed image by using a color histogram of a standard color correction card, so that the color cast problem of the primary processed image caused by the environment is reduced, and a secondary processed image is obtained. For example, in the non-local mean image denoising algorithm, the search region may be 21 × 21, and the similarity comparison block may be 7 × 7. The search area and the comparison block area need to be increased as the standard deviation of gaussian noise is larger. The image color correction algorithm loads a reference image, namely a color correction card, and an image to be corrected, obtains a color histogram of a color matching card in the reference image through a program, and applies the color histogram to the image to be corrected.
S3: and performing data fusion on the secondary processing image and the original data to obtain primary processing data.
Specifically, a certain fixed area in the center of the secondary processed image is cut to obtain a tertiary processed image; calculating L, a and b color values of each pixel in the three processed images by using an OpenCV calculation visual library, and taking the average value as an L, a and b color average value of the three processed images; wherein L represents lightness, a represents a component from green to red, b represents a component from blue to yellow; and fusing the characteristics of L, a and b extracted from the three processed images with the original data by taking time as a corresponding relation to obtain primary processed data.
For example, a fixed region of 1400 × 650 pixels in the center of the secondary image is clipped, and the L, a, b color values of each pixel in the region are sequentially calculated, and the average value thereof is calculated as the average value of the L, a, b color values of the image. In addition, the original image information and the original data information are in one-to-one correspondence in time relationship, so that the extracted L, a, b characteristics and the original data can be fused in the correspondence relationship of time to form primary processing data.
S4: respectively importing the five-time processing data into a decision tree model and a support vector machine model, and fusing prediction results by using a weighted voting method according to the performance of the models;
specifically, fusion data after data preprocessing, namely L, a and b color values, air-curing humidity and cigar tobacco leaf water loss amount are used as input characteristic parameters, and digital codes in the air-curing stage are used as input labels and are led into a decision tree model and a support vector machine model for training and optimizing. And after the prediction accuracy rates of the two models are obtained, comparing, giving higher weight to the model with higher accuracy rate, and finally, synthesizing according to the prediction probability and the weight of the two models to obtain a prediction result.
In one embodiment, S4 specifically includes: s401: according to the corresponding relation between the air-curing time and the air-curing stage in the air-curing process of the cigar tobacco leaves, marking each piece of primary processing data to correspond to the air-curing stage, dividing the primary processing data into a withering stage, a yellowing stage, a browning stage and a drying stage, and converting the primary processing data into digital quantities such as 0, 1, 2 and 3 to obtain secondary processing data; s402: when a small amount of missing characteristic values exist in the secondary processing data, performing interpolation filling by using the mean values of the adjacent characteristic values before and after the small amount of missing characteristic values to obtain tertiary processing data; s403: abnormal value identification is carried out on the three-time processed data by using a standard deviation detection method, and the abnormal value is corrected by using the functional relation of the data before and after the abnormal value, so that four-time processed data are obtained; s404: carrying out data standardization processing on the four processed data by using a Z-Score standardization method to obtain five processed data; s405: and importing the five times of processing data into a decision tree model for training and prediction. Wherein the water loss amount of the tobacco leaves in the air-curing process, the humidity of the air-curing room, L, a and b are used as input characteristics, and the air-curing stage is used as an input label; determining input characteristics by using the information gain ratio in the training process, and constructing a decision tree; taking the obtained classification probability and the classification result as a first result, and marking as R1; s406: and importing the five times of processing data into a support vector machine model for training and prediction. Wherein the water loss amount of the tobacco leaves in the air-curing process, the humidity of the air-curing room, L, a and b are used as input characteristics, and the air-curing stage is used as an input label; the training process uses a grid searching method to automatically adjust model parameters; taking the obtained classification probability and the classification result as a second result, and marking as R2; s407: and carrying out weight division on the models according to the accuracy ranking, testing the prediction accuracy after multi-model fusion under different weights, and selecting the most appropriate weight proportion.
The effects of the present invention will be further described below with reference to the predicted results.
Inputting ten same new data into the decision tree model, counting the prediction result of the model, directly recording the possibility of the model for predicting the new data, and considering that the prediction result is correct when the prediction result is more than 50%. Ten same new data are input into the support vector machine model, the prediction result of the model is counted, and the possibility of the model for predicting the new data is directly recorded. The following is assumed:
it can be found that the accuracy of the prediction of the decision tree model is 80%, and the accuracy of the prediction of the support vector machine model is 80%. Carrying out weighted average on the prediction results of the two models, firstly setting the weight value of the decision tree model to be 0.7, setting the model weight value of the support vector machine to be 0.3, and obtaining the prediction result of the fusion model through weighting as follows:
the accuracy rate is still 80%, but if the weight value of the decision tree model is 0.6, the model weight value of the support vector machine is 0.4, and the prediction result of the fusion model obtained by weighting is as follows:
at this time, the accuracy of the fusion model is 90%, which is better than that of the single model. The weights are given here by way of example only and are not particularly limited.
In the description of the present invention, "a plurality" means two or more unless otherwise specified; the terms "upper", "lower", "left", "right", "inner", "outer", "front", "rear", "head", "tail", and the like, indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, are only for convenience in describing and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed in a particular orientation, and be operated, and thus, should not be construed as limiting the invention. Furthermore, the terms "first," "second," "third," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
It should be noted that the embodiments of the present invention can be realized by hardware, software, or a combination of software and hardware. The hardware portion may be implemented using dedicated logic; the software portions may be stored in a memory and executed by a suitable instruction execution system, such as a microprocessor or specially designed hardware. Those skilled in the art will appreciate that the apparatus and methods described above may be implemented using computer executable instructions and/or embodied in processor control code, such code being provided on a carrier medium such as a disk, CD-or DVD-ROM, programmable memory such as read only memory (firmware), or a data carrier such as an optical or electronic signal carrier, for example. The apparatus and its modules of the present invention may be implemented by hardware circuits such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., or by software executed by various types of processors, or by a combination of hardware circuits and software, e.g., firmware.
The above description is only for the purpose of illustrating the present invention and the appended claims are not to be construed as limiting the scope of the invention, which is intended to cover all modifications, equivalents and improvements that are within the spirit and scope of the invention as defined by the appended claims.
Claims (10)
1. A cigar tobacco leaf process stage prediction method based on data fusion is characterized in that the cigar tobacco leaf process stage prediction method based on data fusion comprises the following steps:
acquiring original data and original images in the tobacco leaf airing process of the cigar in a preset acquisition period; respectively processing original data and original images, training a decision tree model and a support vector machine model based on the processed images and data, and fusing prediction results by using a weighted voting method; and acquiring original data and original images of the cigar tobacco leaves in the airing process in real time, processing the images to be used as model input items, and distinguishing the cigar tobacco leaves in the airing stage.
2. The data fusion-based cigar tobacco processing stage prediction method according to claim 1, wherein the data fusion-based cigar tobacco processing stage prediction method comprises the steps of:
dividing cigars to be aired into different types according to attributes and varieties; when the cigars are aired, acquiring airing room environment data and cigar tobacco leaf data in the airing process in a certain acquisition period to obtain original data and an original image;
step two, performing unstructured data preprocessing on the collected original images of the cigar tobacco leaves to obtain primary processed images and secondary processed images;
step three, cutting the obtained secondary processing image to obtain a tertiary processing image; performing data fusion on the obtained three-time processed image and the original data to obtain primary processed data;
processing the primary processing data to sequentially obtain secondary processing data, tertiary processing data, quartic processing data and quintic processing data;
importing the five-time processing data into a decision tree model and a support vector machine model, and fusing a prediction result by using a weighted voting method according to the performance of the models;
and step six, acquiring original data and original images of the cigar tobacco leaves in the airing process in real time, processing the original data and the original images to be used as input items of a decision tree model and a support vector machine model, and distinguishing the cigar tobacco leaves in the airing stage.
3. The data fusion-based cigar tobacco processing stage prediction method of claim 2, wherein the environmental data includes air-drying humidity; the cigar tobacco leaf data comprises tobacco leaf air-curing images and water loss data.
4. The method for predicting the cigar tobacco processing stage based on data fusion as claimed in claim 2, wherein the collecting the airing environment data and the airing process cigar tobacco data in a certain collection period to obtain the original data and the original image comprises:
collecting an air-cured image of the cigar tobacco leaves as an original image by utilizing a camera in a preset collection period; and collecting the air-curing water loss amount of the cigar tobacco leaves and the humidity data of the air-curing room as original data by utilizing each sensor in a preset collection period.
5. The method of claim 2, wherein the step of collecting the data of the house-airing environment and the data of the cigar tobacco leaves in the airing process in a certain collection period to obtain the raw data and the raw image further comprises:
unifying the format of the collected original data; unifying the format of the collected original images.
6. The data fusion-based cigar tobacco processing stage prediction method of claim 2, wherein the step of performing unstructured data preprocessing on the collected cigar tobacco raw image to obtain a primary processed image and a secondary processed image comprises:
determining pixels with similar neighborhood structures in an original image by using a non-local mean image noise reduction algorithm and taking the pixels as units; then, the determined area is weighted and averaged to obtain a primary processing image;
and performing color matching correction on the primary processed image by using an image color correction algorithm and by using a color histogram of a standard color correction card to obtain a secondary processed image.
7. The data fusion-based cigar tobacco process stage prediction method of claim 2, wherein the obtained secondary processed image is cropped to obtain a tertiary processed image; performing data fusion on the obtained three processed images and the original data to obtain primary processed data, wherein the primary processed data comprises the following steps:
cutting a certain fixed area in the center of the obtained secondary processing image to obtain a tertiary processing image;
calculating L, a and b color values of each pixel in the three processed images by using an OpenCV calculation visual library, and taking the average value as an L, a and b color average value of the three processed images; wherein L represents lightness, a represents a component from green to red, b represents a component from blue to yellow;
and fusing the characteristics of L, a and b extracted from the three processed images with the original data by taking time as a corresponding relation to obtain primary processed data.
8. The data fusion-based cigar tobacco processing stage prediction method according to claim 2, wherein the processing the primary processing data to obtain the secondary processing data, the tertiary processing data, the quaternary processing data and the quintic processing data in sequence comprises:
according to the corresponding relation between the air-curing time and the air-curing stage in the air-curing process of the cigar tobacco leaves, marking each primary processing data with a corresponding air-curing stage, and dividing the primary processing data into a withering stage, a yellowing stage, a browning stage and a tendon drying stage to obtain secondary processing data;
when a small amount of missing characteristic values exist in the secondary processing data, performing interpolation filling by using the mean values of the adjacent characteristic values before and after the small amount of missing characteristic values to obtain tertiary processing data;
abnormal value identification is carried out on the three-time processed data by using a standard deviation detection method, and the abnormal value is corrected by using the functional relation of the data before and after the abnormal value to obtain four-time processed data;
carrying out data standardization processing on the four processed data by using a Z-Score standardization method to obtain five processed data;
the importing the five-time processing data into a decision tree model and a support vector machine model, and fusing prediction results by using a weighted voting method according to the performance of the models comprises the following steps:
importing the five processed data into a decision tree model for training and prediction comprises the following steps:
taking the water loss amount of tobacco leaf in air-curing, the humidity of an air-curing room, L, a and b as input characteristics, and taking the air-curing stage as an input label; determining input characteristics by using the information gain ratio in the training process, and constructing a decision tree; taking the obtained classification probability and the classification result as a first result, and marking as R1;
importing the five processed data into a support vector machine model for training and prediction comprises the following steps:
taking the water loss amount of tobacco leaf in air-curing, the humidity of an air-curing room, L, a and b as input characteristics, and taking the air-curing stage as an input label; the training process uses a grid searching method to automatically adjust model parameters; taking the obtained classification probability and the classification result as a second result, and marking as R2;
and carrying out weight division on the models according to the accuracy ranking, testing the prediction accuracy after multi-model fusion under different weights, and selecting the most appropriate weight proportion.
9. A data fusion-based cigar tobacco processing stage prediction system for implementing the data fusion-based cigar tobacco processing stage prediction method according to any one of claims 1 to 8, wherein the data fusion-based cigar tobacco processing stage prediction system comprises:
the system comprises an original data and original image acquisition module, a data processing module and a data processing module, wherein the original data and original image acquisition module is used for dividing cigars to be aired into different types according to attributes and varieties; when the cigars are aired, acquiring airing room environment data and cigar tobacco leaf data in the airing process in a certain acquisition period to obtain original data and an original image;
the processing image acquisition module is used for carrying out unstructured data preprocessing on the acquired original images of the cigar tobacco leaves to obtain primary processing images and secondary processing images;
the line data fusion module is used for cutting the obtained secondary processing image to obtain a tertiary processing image; performing data fusion on the obtained three-time processed image and the original data to obtain primary processed data;
the multi-processing data acquisition module is used for processing the primary processing data to sequentially obtain secondary processing data, tertiary processing data, quartic processing data and quintic processing data;
the fusion prediction result module is used for importing the five-time processing data into a decision tree model and a support vector machine model and fusing prediction results by using a weighted voting method according to the performance of the models;
and the judging module is used for acquiring original data and original images of the cigar tobacco leaves in the airing process in real time, processing the original images to be used as input items of a decision tree model and a support vector machine model, and judging the cigar tobacco leaves in the airing stage.
10. A computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to carry out the data fusion-based cigar tobacco process stage prediction method according to any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210032426.XA CN114266419A (en) | 2022-01-12 | 2022-01-12 | Cigar tobacco leaf process stage prediction method, system and medium based on data fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210032426.XA CN114266419A (en) | 2022-01-12 | 2022-01-12 | Cigar tobacco leaf process stage prediction method, system and medium based on data fusion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114266419A true CN114266419A (en) | 2022-04-01 |
Family
ID=80832943
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210032426.XA Pending CN114266419A (en) | 2022-01-12 | 2022-01-12 | Cigar tobacco leaf process stage prediction method, system and medium based on data fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114266419A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114860012A (en) * | 2022-07-07 | 2022-08-05 | 湖北省烟草科学研究院 | Intelligent monitoring and regulating system for cigar airing based on Internet of things |
| CN117876380A (en) * | 2024-03-13 | 2024-04-12 | 昆明昊拜农业科技有限公司 | Tobacco leaf environment temperature and humidity and microlayer difference prediction method and system |
-
2022
- 2022-01-12 CN CN202210032426.XA patent/CN114266419A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114860012A (en) * | 2022-07-07 | 2022-08-05 | 湖北省烟草科学研究院 | Intelligent monitoring and regulating system for cigar airing based on Internet of things |
| CN117876380A (en) * | 2024-03-13 | 2024-04-12 | 昆明昊拜农业科技有限公司 | Tobacco leaf environment temperature and humidity and microlayer difference prediction method and system |
| CN117876380B (en) * | 2024-03-13 | 2024-05-14 | 昆明昊拜农业科技有限公司 | Tobacco leaf environment temperature and humidity and microlayer difference prediction method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114266419A (en) | Cigar tobacco leaf process stage prediction method, system and medium based on data fusion | |
| CN108764372A (en) | Construction method and device, mobile terminal, the readable storage medium storing program for executing of data set | |
| CN110751019B (en) | High-resolution image crop automatic extraction method and device based on deep learning | |
| CN108875821A (en) | The training method and device of disaggregated model, mobile terminal, readable storage medium storing program for executing | |
| US20240013380A1 (en) | Cigar tobacco leaf harvesting maturity identification method and system based on integrated learning | |
| CN107704878B (en) | Hyperspectral database semi-automatic establishment method based on deep learning | |
| CN110533654A (en) | The method for detecting abnormality and device of components | |
| CN112200011B (en) | Aeration tank state detection method, system, electronic equipment and storage medium | |
| CN111160096A (en) | Method, device and system for identifying poultry egg abnormality, storage medium and electronic device | |
| CN115909059A (en) | Natural resource sample library establishing method and device | |
| CN114049325A (en) | Construction method and application of lightweight face mask wearing detection model | |
| CN113194359A (en) | Method, device, equipment and medium for automatically grabbing baby wonderful video highlights | |
| CN115292334A (en) | Intelligent planting method and system based on vision, electronic equipment and storage medium | |
| CN115170988A (en) | Power grid line fault identification method and system based on deep learning | |
| CN116563293A (en) | Photovoltaic carrier production quality detection method and system based on machine vision | |
| CN117589767B (en) | Tobacco leaf harvesting time determining method, electronic equipment and storage medium | |
| CN114549513A (en) | Part identification method, part identification device, quality inspection method, electronic device, and storage medium | |
| CN116524725B (en) | Intelligent driving traffic sign image data identification system | |
| CN111178405A (en) | Similar object identification method fusing multiple neural networks | |
| CN110618129A (en) | Automatic power grid wire clamp detection and defect identification method and device | |
| CN110688900A (en) | Withdrawal meter management method based on image recognition | |
| CN116958643A (en) | Air-borne pollen sensitized plant intelligent identification method based on YOLO network | |
| CN116128853A (en) | Production line assembly detection method, system, computer and readable storage medium | |
| CN116311492A (en) | Gesture recognition method and system based on depth camera and contour extraction | |
| CN115809242A (en) | Meteorological data storage method and system |
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 |