CN-122023992-A - Tobacco disease diagnosis method and model based on image processing

CN122023992ACN 122023992 ACN122023992 ACN 122023992ACN-122023992-A

Abstract

The invention discloses a tobacco disease diagnosis method and model based on image processing, which are characterized by collecting original images of field flue-cured tobacco, preprocessing the images, calculating to obtain the incidence probability of each leaf for different diseases K, calculating to obtain weighted average incidence probability of the leaf according to the incidence probability of a plurality of leaves, calculating to obtain the incidence probability of a stalk and the incidence probability of the whole plant, and then carrying out weighted average according to the weighted average incidence probability of the leaf, the incidence probability of the stalk and the incidence probability of the whole plant, and calculating to obtain the incidence probability of the plant for the diseases K. The invention has the advantages that the invention can integrate the image characteristics of the leaves, the stems and the whole plant, realize the cooperative diagnosis of multiple parts, improve the comprehensiveness and the accuracy of disease identification, realize the fusion of multi-source information and effectively improve the disease diagnosis accuracy.

Inventors

XIA TIYUAN
FAN ZHIWEI
WANG YANHONG
ZHONG YU
TANG ZUOXIN
WU CHENG
PAN YIHONG
ZHU XIAOHONG
Ren Die
ZHANG YU

Assignees

昆明学院

Dates

Publication Date: 20260512
Application Date: 20260204

Claims (10)

1. A tobacco disease diagnosis method based on image processing is characterized by comprising the following steps: S1, acquiring original images of field flue-cured tobacco, wherein the original images comprise an integral image of a plant, a stalk image and a plurality of leaf images of the same plant; S2, preprocessing an image, namely preprocessing an original image of the flue-cured tobacco in the acquired field; S3, calculating P leaf-n(k) , namely respectively extracting blade characteristics aiming at each blade image, and calculating the incidence probability of each blade aiming at different diseases K, namely P leaf-n(k) , wherein K represents diseases and leaf-n represents the nth blade of the plant; S4, aggregating leaf evidences, namely calculating to obtain weighted average incidence probability P leaf(k) of the leaves according to P leaf-n(k) of a plurality of leaves; S5, calculating P stem(k) , namely extracting stalk characteristics according to stalk images of plants, and calculating to obtain stalk incidence probability P stem(k) of the plants aiming at different diseases K; S6, calculating P whole(k) , namely extracting integral features according to integral images of plants, and calculating to obtain integral plant incidence probability P whole(k) of the plants aiming at different diseases K; S7, calculating the probability P k of the plant for the disease K according to the calculated P leaf(k) 、P stem(k) 、P whole(k) ; S8, arranging from large to small according to the calculated P K , outputting the probability P k of occurrence of the first diseases and the types of the diseases, and displaying the corresponding image characteristics.
2. The method for diagnosing tobacco diseases based on image processing according to claim 1, wherein the calculation method of P leaf-(k) in the step S4 comprises the following steps: S41, calculating weights Wn of a plurality of blades: , Alesion, n is the area of the nth blade; Aleaf, n is the total area of a plurality of lesions on the nth leaf; Is a constant; S42, calculating P leaf(k) : ; Wherein: n is the total number of leaves analyzed; p leaf-n(k) is the incidence probability of the nth leaf on the disease K; wn is the weight of the nth leaf.
3. The method for diagnosing tobacco diseases based on image processing according to claim 1, wherein the calculation method of P k in the step S7 comprises the following steps: s71, determining weights of the leaf, the stem and the whole body, and respectively marking the weights as w leaf(k) ,w stem(k) ,w whole(k) ; S72, calculating P k : 。
4. The method for diagnosing a tobacco disease based on image processing according to claim 3, wherein the method of step S71 comprises the steps of: S711, presetting an initial weight matrix of a leaf, a stem and a whole plant for different diseases k according to experience by an expert, wherein the initial weight matrix is denoted as w leaf(k)1 ,w stem(k)1 ,w whole(k)1 and satisfies the condition that w leaf(k)1 +w stem(k)1 +w whole(k)1 =1; S712, respectively presetting a threshold range of incidence probability for the leaves, stems and whole plants of each disease k; S713, in the actual diagnosis, respectively comparing the calculated P leaf(k) 、P stem(k) 、P whole(k) with the respective incidence probability threshold range, and if the incidence probability threshold range is exceeded, adjusting an initial weight matrix to obtain a final weight matrix, wherein the adjustment rule is that if the probability value is higher than the upper limit, the symptom of the part is very typical, the weight is increased, and if the probability value is lower than the lower limit, the symptom of the part is not obvious, and the weight is reduced; s714, normalizing the adjusted weights to obtain a final weight matrix w leaf(k) ,w stem(k) ,w whole(k) of the leaf, the stalk and the whole plant.
5. The method for diagnosing tobacco diseases based on image processing according to claim 1, wherein the calculation method of P leaf-n(k) in the step S3 comprises the following steps: s31, calculating a characteristic value fi of the blade characteristic extracted by each blade; S32, normalizing the characteristic value fi of each characteristic i; finorm=(fi-μi)/σi; Wherein fi is an original characteristic value, mu i is the mean value of the characteristic in the training set, sigma i is the standard deviation of the characteristic in the training set, finorm is a standardized characteristic value; s33, determining a weight matrix W (K, i) of image features of different diseases K; S34, calculate a base score S k (image) for each disease K: ; wherein b k represents the prior possibility of disease k without characteristic information; S35, calculating preliminary probability of the image features, namely converting the basic score S k (image) into incidence probability P leaf-n(k) : 。
6. The method for diagnosing a tobacco disease based on image processing according to claim 5, wherein the image feature weight matrix W (k, i) in the step S33 is determined by: W(k,i)=α·Wexpert+(1-α)·Wlearned; Wherein Wexpert is a weight matrix empirically determined by an expert; wlearned is a weight matrix obtained by data-driven learning; alpha is an expert trust coefficient, alpha epsilon [0,1], alpha is manually set and can be timely adjusted according to the running condition.
7. The method for diagnosing a tobacco disease based on image processing according to claim 5, wherein μi and σi in step S32 are updated periodically: μi(new)=θ*μi(old)+(1-θ)*μi(newsample); ; Where θ is a forgetting factor, and the value of θ may be defined manually.
8. The method for diagnosing a tobacco disease based on image processing as recited in claim 5, wherein b k in the step S34 is calculated based on a historical probability of occurrence: ; pkprior represents the historical probability of onset of disease k over time.
9. The method for diagnosing tobacco diseases based on image processing according to any one of claims 5 to 8, wherein the calculation method of the stalk attack probability P stem(k) in the step S5 and the calculation method of the whole plant attack probability P whole(k) in the step S6 are the same as those of P leaf-n(k) , except that the corresponding leaf features are replaced with stalk features and plant features, and the weight matrix of the leaf features is replaced with the weight matrix of the stalk features and plant features.
10. The tobacco disease diagnosis model based on image processing is characterized in that the tobacco disease diagnosis method based on image processing according to any one of claims 1-9 can be realized.

Description

Tobacco disease diagnosis method and model based on image processing Technical Field The invention relates to the technical field of automatic disease identification, in particular to a tobacco disease diagnosis method and model based on image processing. Background At present, the diagnosis of tobacco diseases mainly depends on field observation and experience judgment of agricultural technicians, and the method has the problems of strong subjectivity, low efficiency and high dependence on professionals. With the development of image processing and artificial intelligence technology, studies have been attempted to identify diseases based on leaf images. However, the existing methods are mostly focused on leaf spot recognition, the morphological characteristics of stems and whole plants are ignored, systematic symptoms of diseases cannot be comprehensively reflected, and the diagnosis accuracy is low. Therefore, there is an urgent need for an intelligent diagnosis method and model for tobacco diseases, which can integrate multiple image features and has high diagnosis accuracy. Disclosure of Invention The invention aims to solve the technical problem of providing a tobacco disease diagnosis method and model which can be used for fusing multi-part image characteristics and have high diagnosis accuracy. In order to solve the technical problems, the invention provides a tobacco disease diagnosis method based on image processing, which comprises the following steps: S1, acquiring original images of field flue-cured tobacco, wherein the original images comprise an integral image of a plant, a stalk image and a plurality of leaf images of the same plant; S2, preprocessing an image, namely preprocessing an original image of the flue-cured tobacco in the acquired field; S3, calculating P leaf-n(k), namely respectively extracting blade characteristics aiming at each blade image, and calculating the incidence probability of each blade aiming at different diseases K, namely P leaf-n(k), wherein K represents diseases and leaf-n represents the nth blade of the plant; S4, aggregating leaf evidences, namely calculating to obtain weighted average incidence probability P leaf(k) of the leaves according to P leaf-n(k) of a plurality of leaves; S5, calculating P stem(k), namely extracting stalk characteristics according to stalk images of plants, and calculating to obtain stalk incidence probability P stem(k) of the plants aiming at different diseases K; S6, calculating P whole(k), namely extracting integral features according to integral images of plants, and calculating to obtain integral plant incidence probability P whole(k) of the plants aiming at different diseases K; S7, calculating the probability P k of the plant for the disease K according to the calculated P leaf(k)、Pstem(k)、Pwhole(k); S8, arranging from large to small according to the calculated P K, outputting the probability P k of occurrence of the first diseases and the types of the diseases, and displaying the corresponding image characteristics. Further, the calculation method of P leaf-(k) in step S4 includes the following steps: S41, calculating weights Wn of a plurality of blades: Alesion, n is the area of the nth blade; Aleaf, n is the total area of a plurality of lesions on the nth leaf; Is a constant; S42, calculating P leaf(k): ; Wherein: n is the total number of leaves analyzed; p leaf-n(k) is the incidence probability of the nth leaf on the disease K; wn is the weight of the nth leaf. Further, the calculation method of P k in step S7 includes the following steps: s71, determining weights of the leaf, the stem and the whole body, and respectively marking the weights as w leaf(k),wstem(k),wwhole(k); S72, calculating P k: 。 further, the method of step S71 includes the following steps: S711, presetting an initial weight matrix of a leaf, a stem and a whole plant for different diseases k according to experience by an expert, wherein the initial weight matrix is denoted as w leaf(k)1,wstem(k)1,wwhole(k)1 and satisfies the condition that w leaf(k)1+wstem(k)1+wwhole(k)1 =1; S712, respectively presetting a threshold range of incidence probability for the leaves, stems and whole plants of each disease k; S713, in the actual diagnosis, respectively comparing the calculated P leaf(k)、Pstem(k)、Pwhole(k) with the respective incidence probability threshold range, and if the incidence probability threshold range is exceeded, adjusting an initial weight matrix to obtain a final weight matrix, wherein the adjustment rule is that if the probability value is higher than the upper limit, the symptom of the part is very typical, the weight is increased, and if the probability value is lower than the lower limit, the symptom of the part is not obvious, and the weight is reduced; s714, normalizing the adjusted weights to obtain a final weight matrix w leaf(k),wstem(k),wwhole(k) of the leaf, the stalk and the whole plant. Further, the method comprises the steps of,In the ra