CN111860639B - System and method for judging quantized flue-cured tobacco leaf curing characteristics - Google Patents
System and method for judging quantized flue-cured tobacco leaf curing characteristics Download PDFInfo
- Publication number
- CN111860639B CN111860639B CN202010692507.3A CN202010692507A CN111860639B CN 111860639 B CN111860639 B CN 111860639B CN 202010692507 A CN202010692507 A CN 202010692507A CN 111860639 B CN111860639 B CN 111860639B
- Authority
- CN
- China
- Prior art keywords
- module
- tobacco leaves
- follows
- segmentation
- formula
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24147—Distances to closest patterns, e.g. nearest neighbour classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
Abstract
The invention discloses a system for judging the curing characteristics of quantized flue-cured tobacco leaves, which comprises a color segmentation module, wherein the color segmentation module is connected with a uniform illumination module, an image acquisition module, a pre-training module and a data statistics module, and the data statistics module is connected with an output module. Has the following advantages: and (4) quantifying the yellowing and browning areas of the fresh tobacco leaves, and accurately judging the baking characteristics of the fresh tobacco leaves.
Description
Technical Field
The invention belongs to the technical field of flue-cured tobacco modulation, and particularly relates to a system and a method for judging the curing characteristics of quantized flue-cured tobacco leaves.
Background
The tobacco leaf roasting characteristics refer to the yellowing and dehydration speed and synchronization degree of the tobacco leaves in the roasting process, whether the tobacco leaves are easy to color after yellowing and other characteristics, including easy roasting performance and roasting resistance. Fresh tobacco leaves which are easy to turn yellow and have good yellow and dehydration coordination have good easy-to-bake property, and conversely, the easy-to-bake property is poor. The longer the leaf turns yellow and remains non-browning, the better its bake resistance and the better the bake characteristics. The baking characteristics of tobacco leaves have a very close relationship with the quality of the tobacco leaves after baking, because the fresh tobacco leaves harvested in the field can be cured to show and fix the excellent quality of the tobacco leaves and become the commercial tobacco leaves.
At home and abroad, tobacco technologists pay attention to the research on the baking characteristics of tobacco leaves. At present, most of researches on the baking characteristics of tobacco leaves are focused on the changes of moisture content, pigment content and color parameters, and quantitative indexes related to the baking characteristics are mainly judged by the yellowing time and the browning time of fresh tobacco leaves in a dark box test. The yellowing time represents the easy-to-bake property of the tobacco leaves, and is based on the time required for the tobacco leaves to reach full yellow. The shorter the yellowing time, the better the bake-ability, and conversely the lower the bake-ability. The browning time represents the baking resistance of the tobacco leaves, and is based on the time from full yellowing of the tobacco leaves to three browning of the tobacco leaves (the browning area accounts for 30% of the whole tobacco leaf area). The longer the browning time, the better the baking resistance, whereas the worse the baking resistance.
CN102172296A discloses a method for judging the baking characteristics of flue-cured tobacco, when the yellowing time of lower leaves is 48-60h, and the middle and upper leaves are about 72h, the easy baking is better; when the browning time of the lower leaves is more than 72h, the browning time of the middle leaves is more than 120h, and the browning time of the upper leaves is more than 60h, the roasting resistance is better. However, the disadvantage of this patent is that the areas of yellowing and browning of the tobacco leaves are judged by manual experience.
At present, researches find that the color parameters of the fresh tobacco leaves can be used as quantitative indexes for judging the baking characteristics of the fresh tobacco leaves, but no clear quantitative method is provided.
Disclosure of Invention
Aiming at the defects, the invention provides a system and a method for judging the baking characteristics of the flue-cured tobacco leaves quantitatively, so as to quantify the yellowing and browning areas of the fresh tobacco leaves and accurately judge the baking characteristics of the fresh tobacco leaves.
In order to solve the technical problems, the invention adopts the following technical scheme:
a judging system for quantifying flue-cured tobacco leaf curing characteristics comprises a color segmentation module, wherein the color segmentation module is connected with a uniform illumination module, an image acquisition module, a pre-training module and a data statistics module, and the data statistics module is connected with an output module;
the uniform illumination module is an illumination system, so that the tobacco leaves can be uniformly illuminated, and the subsequent segmentation treatment is facilitated;
the pre-training module is used for carrying out pre-manual segmentation on the sample image in an HSV color space and training a KNN classifier or a SVM classifier;
the image acquisition module is used for acquiring an integral image of the tobacco leaves in an off-line or on-line manner;
the color segmentation module is used for segmenting the tobacco leaf image through a pre-trained KNN classifier or a Support Vector Machine (SVM) classifier to obtain a corresponding result;
the data statistics module is used for counting the areas of the tobacco leaves and the areas of various segmentation colors and carrying out data calculation statistics on various parameters;
the output module is used for displaying or outputting the acquired various data.
A method for quantifying cured tobacco leaf baking characteristics, the method comprising the steps of:
step 1, collecting an image of fresh tobacco leaves;
step 2, carrying out color difference elimination on the collected fresh tobacco leaf images by adopting a CIE-lab color difference formula;
step 3, detecting the picture by adopting a maximum stable value region MSER and a maximum inter-class variance method Otsu to perform integral blade segmentation;
step 4, the color segmentation module and the pre-training module perform color space segmentation and training by utilizing a hexagonal cone model HSV and a proximity algorithm KNN or a support vector machine SVM classifier;
and 5, finally, outputting the segmentation picture, the yellow area ratio and the brown area ratio by the data statistics module and the output module so as to judge the baking characteristics of the tobacco leaves.
Further, the specific method for acquiring the image of the fresh tobacco leaf in the step 1 is as follows:
when tobacco leaves at each part of flue-cured tobacco are mature, picking up the mature tobacco leaves in the field, placing the tobacco leaves in a dark, non-ventilated and room-temperature environment for a dark box test, then shooting images of the fresh tobacco leaves every 12 hours by using an image acquisition module, flatly paving the fresh tobacco leaves on a black matte plate during shooting, taking a standard light source in an even illumination module as a light background, and shooting the fresh tobacco leaves at a position which is vertically 0.5m away from the image acquisition module.
Further, the CIE-lab color difference formula in step 2 is as follows:
further, the maximum stable value region (MSER) calculation formula in step 3 is as follows:
wherein Q is i Denotes the area of the ith connected region, Δ is a small change in the grayscale threshold, and when vi is smaller than a given threshold, this region is considered to be MESR.
Further, the calculation formula of the maximum inter-class variance method otsu in step 3 is specifically as follows:
let T be the segmentation threshold of the foreground and the background, and T divides the image into the foreground and the background. The ratio of foreground points to image is w 0 Average gray of u 0 (ii) a The number of background points in the image is w 1 Average gray level of u 1 Then the total average gray scale of the image is: u-w 0 *u 0 +w 1 *u 1 ；
The variance of the foreground and background images is calculated as follows:
g＝w 0 *(u 0 -u)*(u 0 -u)+w 1 *(u 1 -u)*(u 1 -u)＝w 0 *w 1 *(u 0 -u 1 )*(u 0 -u 1 )；
when the variance g is maximum, it can be considered that the difference between the foreground and the background is maximum at this time, that is, the gray level at this time is the optimal threshold.
Further, the specific method of the proximity algorithm in step 4 is as follows:
if most of k nearest samples of a sample in a feature space belong to a certain class, the sample is also divided into the class, in the KNN algorithm, the selected nearest classes are all objects which are classified correctly, and the method only determines the class of the sample to be classified according to the class of the nearest sample or samples in the classification decision;
first to L P The distance is defined as follows:
wherein x i ∈R n ，x j ∈R n Wherein L ∞ is defined as:
where P is a variable, and when P is 2, L P The distance becomes the euclidean distance corresponding to the L2 norm, which is used when selecting two instance similarities.
A representation of the distance between two points or between multiple points, also known as the euclidean metric, defined in euclidean space as the euclidean distance between two points x1(x11, x12, …, x1n) and x2(x21, x22, …, x2n) in euclidean space;
the calculation formula of the Euclidean distance is as follows:
the L2 norm is expressed as:
L 2 is defined as
Wherein
Further, the basic model of the SVM classifier in the step 4 is established as follows:
finding a maximum interval, wherein omega is a normal vector and is perpendicular to the hyperplane, the direction of the hyperplane is determined, b is an offset term, the distance between the hyperplane and an origin is determined, and the sum gamma of the distances from the two heterogeneous support vectors to the hyperplane is:
the partition hyperplane with the maximum interval is found in advance, parameters omega and b are found to enable gamma to be maximum, and the specific formula is as follows:
s.t.y i (ω T x i +b)≥1，i＝1，2，…，m.；
the numerator of the interval formula is constant, so that the interval is maximized, and only the minimum denominator | ω | is needed;
s.t.y i (ω T x i +b)≥1，i＝1，2，…，m.；
for the sake of simplicity, the above formula converts the minimization denominator into the minimization | ω | | for simplicity 2 ；
The Lagrange multiplier method is adopted to solve the dual problem:
the first step is as follows: introducing lagrange multiplier alpha i The Lagrangian function is obtained when the value is more than or equal to 0, and the specific formula is as follows:
the second step is that: let L (ω, b, α) have zero partial derivatives for ω and b, and the specific formula is as follows:
the third step: the concrete formula is as follows:
the final model has the following specific formula:
the KKT condition has the following specific formula:
sparsity of support vector machine solution: after training is completed, most training samples do not need to be reserved, and the final model is only related to the support vectors.
By adopting the technical scheme, compared with the prior art, the invention has the following technical effects:
the yellowing area and the browning time of the fresh tobacco leaves during the dark box test can be quantified, so that the artificial subjectivity is avoided, and the accuracy and the objectivity of data are enhanced;
secondly, this patent technique also can be used for reference in the application of judging fresh tobacco leaf maturity.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below. Throughout the drawings, like elements or portions are generally identified by like reference numerals. In the drawings, elements or portions are not necessarily drawn to scale.
FIG. 1 is a schematic diagram of the determination method of the present invention;
FIG. 2 is a diagram illustrating the color segmentation and classification effects of the determination method of the present invention;
FIG. 3 is a diagram showing the results of the determination according to the present invention;
fig. 4 is a block diagram showing the structure of the determination system of the present invention.
Detailed Description
Embodiment 1, as shown in fig. 4, a system for determining the curing characteristics of a quantized flue-cured tobacco leaf comprises a color segmentation module, wherein the color segmentation module is connected with an even illumination module, an image acquisition module, a pre-training module and a data statistics module, and the data statistics module is connected with an output module.
The uniform illumination module is an illumination system, so that the tobacco leaves can be uniformly illuminated, and the subsequent segmentation treatment is facilitated.
The pre-training module is used for carrying out pre-manual segmentation on the sample image in an HSV color space and training a KNN classifier or a SVM classifier.
The image acquisition module is used for acquiring the whole image of the tobacco leaves in an off-line or on-line manner.
The color segmentation module is used for segmenting the acquired images through a pre-trained KNN classifier or a Support Vector Machine (SVM) classifier to obtain corresponding results, such as green, yellow, brown and the like.
The data statistics module is used for counting the areas of the tobacco leaves and the areas of various segmentation colors and carrying out data calculation statistics on various parameters.
The output module is used for displaying or outputting the acquired various data.
The method for judging the flue-cured tobacco leaf curing characteristics comprises the following steps:
step 1, collecting an image of fresh tobacco leaves;
the experimental tobacco leaves are selected from Yunyan 85 with excellent baking resistance and big white tendon 599 with poor baking resistance which are planted in Qingdao test base of tobacco institute of Chinese academy of agricultural sciences in 2018;
when tobacco leaves at each part of flue-cured tobacco are mature, picking up the mature tobacco leaves in the field, placing the tobacco leaves in a dark, non-ventilated and room-temperature environment for a dark box test, then shooting images of the fresh tobacco leaves every 12 hours by using an image acquisition module, flatly paving the fresh tobacco leaves on a black matte plate during shooting, taking a standard light source in an even illumination module as a light background, and shooting the fresh tobacco leaves at a position which is vertically 0.5m away from the image acquisition module.
Step 2, carrying out color difference elimination on the collected fresh tobacco leaf images by adopting a CIE-lab color difference formula, wherein the CIE-lab color difference formula is as follows:
and 3, detecting the picture by adopting a maximum stable value region (MSER) and a maximum inter-class variance method (Otsu), and performing integral segmentation on the blade.
The maximum stable value region (MSER) calculation formula is as follows:
wherein Q is i Denotes the area of the ith connected region, Δ is a small change in the grayscale threshold, and when vi is smaller than a given threshold, this region is considered to be MESR.
The calculation formula of the maximum inter-class variance method otsu in the step 3 is specifically as follows:
let T be the segmentation threshold of the foreground and the background, and T divides the image into the foreground and the background. The ratio of the foreground points to the image is w 0 Average gray level of u 0 (ii) a The number of background points in the image is w 1 Average gray of u 1 Then the total average gray level of the image is: u-w 0 *u 0 +w 1 *u 1 ；
The variance of the foreground and background images is calculated as follows:
g＝w 0 *(u 0 -u)*(u 0 -u)+w 1 *(u 1 -u)*(u 1 -u)＝w 0 *w 1 *(u 0 -u 1 )*(u 0 -u 1 )；
when the variance g is maximum, it can be considered that the difference between the foreground and the background is maximum at this time, that is, the gray level at this time is the optimal threshold.
And 4, performing color space segmentation and training by using a hexagonal cone model HSV and a proximity algorithm KNN or a support vector machine SVM classifier by using a color segmentation module and a pre-training module, and selecting 33 fresh tobacco leaves with different yellowing or browning to test the effect, wherein the specific effect is shown in figure 2.
The proximity algorithm KNN is specifically as follows:
if most of k nearest samples of a sample in a feature space belong to a certain class, the sample is also divided into the class, all selected neighbors in the KNN algorithm are objects which are classified correctly, and the method only determines the class of the sample to be classified according to the class of the nearest sample or samples in the classification decision.
First to L P The distance is defined as follows:
wherein x is i ∈R n ，x j ∈R n Wherein L ∞ is defined as:
where P is a variable, when P is 1, L P The distance becomes the manhattan distance corresponding to the norm L1, i.e. the sum of the distances of the projections of the line segment formed by two points on the fixed rectangular coordinate system of euclidean space to the axis, e.g. the manhattan distance of point P1 at coordinates (x1, y1) and point P2 at coordinates (x2, y2) on a plane is: | x 1 -x 2 ∣+∣y 1 -y 2 | it is noted that manhattan distance depends on the degree of rotation of the coordinate system, not the translation or mapping of the system on the coordinate axes.
The Manhattan distance calculation formula is as follows:
the L1 norm is expressed as:
L 1 is defined as
Wherein
When P is 2, LP distance becomes euclidean distance corresponding to the norm L2, which is commonly used when selecting two example similarities.
The most common representation of the distance between two points or between multiple points, also known as the euclidean metric, is defined in euclidean space as the euclidean distance between two points x1(x11, x12, …, x1n) and x2(x21, x22, …, x2n) in euclidean space.
The calculation formula of the Euclidean distance is as follows:
the L2 norm is expressed as:
L 2 is defined as
Wherein
When p → ∞, the LP distance becomes the chebyshev distance.
Chebyshev distance between two points a (x1, y1) and b (x2, y2) in two-dimensional plane:
d 12 ＝max(∣x 1 -x 2 ∣，∣y 1 -y 2 ∣)
chebyshev distances of N-dimensional space points a (x11, x12, …, x1N) and b (x21, x22, … x 2N):
d 12 ＝max(∣x 1i -x 2i ∣
the basic model of the SVM classifier is established as follows:
finding a maximum interval, wherein omega is a normal vector and is perpendicular to the hyperplane, the direction of the hyperplane is determined, b is an offset term, the distance between the hyperplane and an origin is determined, and the sum gamma of the distances from the two heterogeneous support vectors to the hyperplane is:
the partition hyperplane with the maximum interval is found in advance, parameters omega and b are found to enable gamma to be maximum, and the specific formula is as follows:
s.t.y i (ω T x i +b)≥1，i＝1，2，…，m.
the numerator of the above interval formula is constant, so that the interval is maximized, only the minimum denominator | ω | is needed.
s.t.y i (ω T x i +b)≥1，i＝1，2，…，m.
For the sake of simplicity, the above formula converts the minimization denominator into the minimization | ω | | for simplicity 2 。
The Lagrange multiplier method is adopted to solve the dual problem:
the first step is as follows: introducing lagrange multiplier alpha i The Lagrangian function is obtained when the value is more than or equal to 0, and the specific formula is as follows:
the second step is that: let L (ω, b, α) have zero partial derivatives for ω and b, and the specific formula is as follows:
the third step: the concrete formula is as follows:
the final model has the following specific formula:
the KKT condition has the following specific formula:
sparsity of support vector machine solution: after training is completed, most training samples do not need to be reserved, and the final model is only related to the support vectors.
And 5, finally, outputting the segmentation picture, the yellow area ratio and the brown area ratio by the data statistics module and the output module so as to judge the baking characteristics of the tobacco leaves, wherein the specific result is shown in the table 1, and the effect graph is shown in the figure 3.
Table 1:
TABLE 1 quantification of the method of the invention
The time and the yellowing/browning area of the fresh tobacco leaves are combined, the baking characteristics of the fresh tobacco leaves of the variety Yunyan 85 and the big white rib 599 are judged according to YC/T311-2009 (national tobacco agency), the result is that the variety Yunyan 85 has better baking resistance and the big white rib has poorer baking resistance, the quality characteristics of the baked tobacco are subjected to appearance evaluation and classification, and the appearance evaluation and classification are found to be matched with the judged baking characteristics.
The description of the present invention has been presented for purposes of illustration and description, and is not intended to be exhaustive or limited to the invention in the form disclosed. Many modifications and variations will be apparent to practitioners skilled in this art. The embodiment was chosen and described in order to best explain the principles of the invention and the practical application, and to enable others of ordinary skill in the art to understand the invention for various embodiments with various modifications as are suited to the particular use contemplated.
Claims (1)
1. A method for judging the curing characteristics of quantized flue-cured tobacco leaves is characterized by comprising the following steps: the judging method is applied to a judging system for quantifying the baking characteristics of the flue-cured tobacco leaves, the judging system comprises a color segmentation module, the color segmentation module is connected with a uniform illumination module, an image acquisition module, a pre-training module and a data statistics module, and the data statistics module is connected with an output module;
the uniform illumination module is an illumination system, so that the tobacco leaves can be uniformly illuminated, and the subsequent segmentation processing is facilitated;
the pre-training module is used for carrying out pre-manual segmentation on the sample image in an HSV color space and training a KNN classifier or a SVM classifier;
the image acquisition module is used for acquiring an integral image of the tobacco leaves in an off-line or on-line manner;
the color segmentation module is used for segmenting the tobacco leaf image through a pre-trained KNN classifier or a Support Vector Machine (SVM) classifier to obtain a corresponding result;
the data statistics module is used for counting the areas of the tobacco leaves and the areas of various segmentation colors and carrying out data calculation statistics on various parameters;
the output module is used for displaying or outputting the acquired various data;
the determination method includes the steps of:
step 1, collecting an image of fresh tobacco leaves;
when tobacco leaves at each part of flue-cured tobacco are mature, picking up field mature tobacco leaves, placing the tobacco leaves in a dark, non-ventilated and room-temperature environment for a dark box test, then shooting images of the fresh tobacco leaves every 12 hours by using an image acquisition module, paving the fresh tobacco leaves on a black matte plate during shooting, taking a standard light source in a uniform illumination module as a light background, and shooting the fresh tobacco leaves at a position which is vertically 0.5m away from the image acquisition module;
step 2, carrying out color difference elimination on the collected fresh tobacco leaf images by adopting a CIE-lab color difference formula;
the CIE-lab color difference formula in the step 2 is as follows:
step 3, detecting the picture by adopting a maximum stable value region MSER and a maximum inter-class variance method Otsu to perform integral blade segmentation;
the maximum stable value region (MSER) calculation formula is as follows:
wherein Q is i The area of the ith connected region is represented, delta is the tiny change amount of the gray threshold, and when vi is smaller than a given threshold, the region is considered to be MESR;
the calculation formula of the maximum inter-class variance method otsu in the step 3 is specifically as follows:
t is a segmentation threshold of the foreground and the background, the image is divided into the foreground and the background by T, and the number of foreground points accounts for the image in the proportion of w 0 Average gray of u 0 (ii) a The number of background points in the image is w 1 Average gray of u 1 Then the total average gray level of the image is: u-w 0 *u 0 +w 1 *u 1 ；
The variance of the foreground and background images is calculated as follows:
g＝w 0 *(u 0 -u)*(u 0 -u)+w 1 *(u 1 -u)*(u 1 -u)＝w 0 *w 1 *(u 0 -u 1 )*(u 0 -u 1 )；
when the variance g is maximum, the difference between the foreground and the background at this time can be considered to be maximum, that is, the gray level at this time is the optimal threshold;
step 4, the color segmentation module and the pre-training module perform color space segmentation and training by utilizing a hexagonal cone model HSV and a proximity algorithm KNN or a support vector machine SVM classifier;
the specific method of the proximity algorithm in the step 4 is as follows:
if most of k nearest samples of a sample in a feature space belong to a certain class, the sample is also divided into the class, in the KNN algorithm, the selected nearest classes are all objects which are classified correctly, and the method only determines the class of the sample to be classified according to the class of the nearest sample or samples in the classification decision;
the LP distance is defined first:
wherein x i ∈R n ，x j ∈R n Wherein L ∞ is defined as:
where P is a variable, and when P is 2, L P The distance becomes an Euclidean distance, the Euclidean distance corresponds to the L2 norm, and the Euclidean distance is used when the similarity of two examples is selected;
a representation of the distance between two points or between multiple points, also known as the euclidean metric, defined in euclidean space as the euclidean distance between two points x1(x11, x12, …, x1n) and x2(x21, x22, …, x2n) in euclidean space;
the calculation formula of the Euclidean distance is as follows:
the L2 norm is expressed as:
L 2 is defined as
Wherein
The basic model of the SVM classifier in the step 4 is established as follows:
finding a maximum interval, wherein omega is a normal vector and is perpendicular to the hyperplane, the direction of the hyperplane is determined, b is an offset term, the distance between the hyperplane and an origin is determined, and the sum gamma of the distances from the two heterogeneous support vectors to the hyperplane is:
the partition hyperplane with the maximum interval is found in advance, parameters omega and b are found to enable gamma to be maximum, and the specific formula is as follows:
s.t.y i (ω T x i +b)≥1，i＝1，2，…，m.；
the numerator of the interval formula is constant, so that the interval is maximized, and only the minimum denominator | ω | is needed;
s.t.y i (ω T x i +b)≥1，i＝1,2，…，m.；
for the sake of simplicity, the above formula converts the minimized denominator into the minimization | ω | | for simplicity 2 ；
The Lagrange multiplier method is adopted to solve the dual problem:
the first step is as follows: introducing lagrange multiplier alpha i The Lagrange function is obtained when the value is more than or equal to 0, and the specific formula is as follows:
the second step is that: let L (ω, b, α) have zero partial derivatives of ω and b, and the following formula is used:
the third step: the concrete formula is as follows:
the final model has the following specific formula:
the KKT condition has the following specific formula:
sparsity of support vector machine solution: after training is finished, most training samples do not need to be reserved, and the final model is only related to the support vector;
and 5, finally, outputting the segmentation picture, the yellow area ratio and the brown area ratio by the data statistics module and the output module so as to judge the baking characteristics of the tobacco leaves.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010692507.3A CN111860639B (en) | 2020-07-17 | 2020-07-17 | System and method for judging quantized flue-cured tobacco leaf curing characteristics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010692507.3A CN111860639B (en) | 2020-07-17 | 2020-07-17 | System and method for judging quantized flue-cured tobacco leaf curing characteristics |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111860639A CN111860639A (en) | 2020-10-30 |
| CN111860639B true CN111860639B (en) | 2022-09-27 |
Family
ID=73001182
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010692507.3A Active CN111860639B (en) | 2020-07-17 | 2020-07-17 | System and method for judging quantized flue-cured tobacco leaf curing characteristics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111860639B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112560692B (en) * | 2020-12-17 | 2023-06-02 | 华侨大学 | Needle mushroom classification system and method based on deep learning |
| CN112568483A (en) * | 2021-01-04 | 2021-03-30 | 中国农业科学院烟草研究所 | Intelligent dark box capable of automatically judging tobacco leaf baking characteristics and judging method |
| CN112801300A (en) * | 2021-01-27 | 2021-05-14 | 福建中烟工业有限责任公司 | Method, device and computer readable medium for predicting aroma amount of tobacco sample |
| CN113919442B (en) * | 2021-02-24 | 2022-05-27 | 北京优创新港科技股份有限公司 | Tobacco maturity state identification method based on convolutional neural network |
| CN113298341A (en) * | 2021-03-01 | 2021-08-24 | 中国烟草总公司郑州烟草研究院 | Method for forecasting appearance color depth of flue-cured tobacco leaves |
| CN113516617B (en) * | 2021-04-02 | 2023-05-05 | 云南省烟草质量监督检测站 | Flue-cured tobacco grade identification modeling method based on machine vision and AI deep learning |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103323455A (en) * | 2013-04-24 | 2013-09-25 | 南京文采科技有限责任公司 | Tobacco leaf grading method based on reflection, perspective and microscopic images |
| CN103919258A (en) * | 2013-03-02 | 2014-07-16 | 重庆大学 | Densification tobacco flue-cure dry-wet bulb temperature automatic control technique based on tobacco image processing |
| CN207603813U (en) * | 2017-12-23 | 2018-07-10 | 四川中德森系统集成有限公司 | A kind of image capturing system for tobacco flue-curing house |
| CN110415181A (en) * | 2019-06-12 | 2019-11-05 | 勤耕仁现代农业科技发展(淮安)有限责任公司 | Flue-cured tobacco RGB image intelligent recognition and grade determination method under a kind of open environment |
-
2020
- 2020-07-17 CN CN202010692507.3A patent/CN111860639B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103919258A (en) * | 2013-03-02 | 2014-07-16 | 重庆大学 | Densification tobacco flue-cure dry-wet bulb temperature automatic control technique based on tobacco image processing |
| CN103323455A (en) * | 2013-04-24 | 2013-09-25 | 南京文采科技有限责任公司 | Tobacco leaf grading method based on reflection, perspective and microscopic images |
| CN207603813U (en) * | 2017-12-23 | 2018-07-10 | 四川中德森系统集成有限公司 | A kind of image capturing system for tobacco flue-curing house |
| CN110415181A (en) * | 2019-06-12 | 2019-11-05 | 勤耕仁现代农业科技发展(淮安)有限责任公司 | Flue-cured tobacco RGB image intelligent recognition and grade determination method under a kind of open environment |
Non-Patent Citations (3)
| Title |
|---|
| 一种基于支持向量机的棉花图像分割算法;陈钦政 等;《计算机工程》;20130531;第39卷(第5期);摘要，第2-3节 * |
| 乙烯利浓度和喷施时期对上部烟叶耐烤性的影响及生理研究;张进 等;《山地农业生物学报》;20200331;第39卷(第2期);第1.2节 * |
| 基于MSER-Otsu与直线矫正的仪表指针定位;秦轩 等;《计算机工程》;20200702;第1.2节 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111860639A (en) | 2020-10-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111860639B (en) | System and method for judging quantized flue-cured tobacco leaf curing characteristics | |
| CN107909081B (en) | Method for quickly acquiring and quickly calibrating image data set in deep learning | |
| CN108734108B (en) | Crack tongue identification method based on SSD network | |
| CN112818827B (en) | Method for judging stage temperature control point in tobacco leaf baking process based on image recognition | |
| CN101059425A (en) | Method and device for identifying different variety green tea based on multiple spectrum image texture analysis | |
| CN104598886B (en) | A kind of method that utilization near-infrared high spectrum image recognizes the oil crops that go mouldy | |
| CN115994907B (en) | Intelligent processing system and method for comprehensive information of food detection mechanism | |
| CN111738931A (en) | Shadow removal algorithm for aerial image of photovoltaic array unmanned aerial vehicle | |
| CN116664565A (en) | Hidden crack detection method and system for photovoltaic solar cell | |
| CN109241932B (en) | Thermal infrared human body action identification method based on motion variance map phase characteristics | |
| CN116434206A (en) | Cotton quality character detection method based on machine vision technology | |
| CN110188693B (en) | Improved complex environment vehicle feature extraction and parking discrimination method | |
| CN110232660B (en) | Novel infrared image recognition preprocessing gray stretching method | |
| CN111563536B (en) | Bamboo strip color self-adaptive classification method based on machine learning | |
| CN116524224A (en) | Machine vision-based method and system for detecting type of cured tobacco leaves | |
| CN115496816A (en) | Sequence clothing image theme color self-adaptive extraction method | |
| CN111832569B (en) | Wall painting pigment layer falling disease labeling method based on hyperspectral classification and segmentation | |
| CN107944426B (en) | Wheat leaf powdery mildew spot marking method based on combination of texture filtering and two-dimensional spectral feature space discrimination | |
| CN109726641B (en) | Remote sensing image cyclic classification method based on automatic optimization of training samples | |
| CN112541859B (en) | Illumination self-adaptive face image enhancement method | |
| CN109918783B (en) | Intelligent clothing design system | |
| CN209842652U (en) | Soybean appearance quality detection device | |
| CN112183633A (en) | Hyperspectral image salient target detection method based on improved FT algorithm | |
| CN116152230B (en) | Textile surface dyeing quality detection method based on spectrum data | |
| CN115841594B (en) | Attention mechanism-based coal gangue hyperspectral variable image domain data identification method |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |