CN112668404B - Effective identification method for soybean diseases and insect pests - Google Patents
Effective identification method for soybean diseases and insect pests Download PDFInfo
- Publication number
- CN112668404B CN112668404B CN202011455125.5A CN202011455125A CN112668404B CN 112668404 B CN112668404 B CN 112668404B CN 202011455125 A CN202011455125 A CN 202011455125A CN 112668404 B CN112668404 B CN 112668404B
- Authority
- CN
- China
- Prior art keywords
- plant
- diseases
- soybean
- insect pests
- value
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture
Landscapes
- Image Processing (AREA)
- Catching Or Destruction (AREA)
Abstract
The invention discloses an effective identification method for soybean diseases and insect pests, which solves the problems of hysteresis and inaccuracy in the aspects of soybean disease and pest identification and identification in the prior art based on field visual observation, laboratory analysis and the like. Establishing a database by utilizing known soybean plant characteristic maps under the influence of different types of soybean diseases and insect pests, extracting and storing various identification parameters related to the soybean plants under the influence of each disease and pest in the database at different growth periods, endowing different identification characteristic parameters with different weights according to the difference of the types of the diseases and the difference of the influences on the soybean plants, and establishing a distinguishing model of the types of the diseases and pests according to the extraction result of the characteristic parameters; the method and the device have the advantages of real-time performance and quasi-real-time performance.
Description
Technical Field
The invention relates to an effective identification method of soybean diseases and insect pests.
Background
The soybean is used as a main grain crop and a commercial crop in China, and is very important in national economic production, particularly agricultural production. Soybean diseases and insect pests are not only in China but also one of the most major agricultural disasters worldwide, and at present, more than 120 kinds of diseases and insect pests related to soybeans are reported in the world, including different types such as fungal diseases, bacterial diseases, viral diseases, insect diseases, parasitic plant diseases and the like, and the modes, the types and the influences of different soybean diseases and insect pests are different, but the growth of the soybeans is seriously influenced by different diseases and insect pests, so that the yield of the soybeans is greatly reduced, and even the soybeans are in an outmost state. High-degree and large-range plant diseases and insect pests cause huge loss to national economy of China and even countries all over the world, particularly national economy of countries mainly based on production and planting of soybean agriculture. The pest and disease damage of the soybean specifically means that the yield of crops suffering from the pest and disease damage is reduced by 1 to 2 on average, and the crops are seriously even dead; after suffering from diseases and insect pests, the taste and quality of crops are reduced to different degrees, and the content of starch and protein is greatly reduced; meanwhile, the soybeans affected by certain diseases and insect pests can cause poisoning, illness and even death of people and animals; in addition, because the diseases and pests generally have larger spreading performance, the large-scale planting of the area cannot be continued due to the influence of the diseases and pests in the same area; in addition, soybeans that are affected by pests are also generally difficult to store and transport, are more susceptible to spoilage than healthy soybeans, and are more susceptible to export restrictions with dangerous pest species.
The effects of pests and diseases on soybeans are mainly reflected on rhizomes, leaves and fruits. For soybean roots, the existence of plant diseases and insect pests can cause the soybean to have root formation, black roots and rotten roots, and the number of root nodules is reduced, so that the absorption of crop roots on water, organic matters and other nutrient substances is influenced, and soybean plants are withered, yellow and slow in growth; for the stems of soybeans, the existence of pests and diseases can cause local disease spots such as anthrax, ulcer and the like to appear on the stems, so that the transportation of moisture and nutrients by the stems of the soybeans is influenced, and further, the soybean plants are withered, rotten, lodging, wilting and even dead; for soybean leaves, the existence of plant diseases and insect pests can cause the color change phenomena such as chlorosis, yellowing and purple change, the color fading phenomena, the spot diseases such as round spots, angular spots and stripes and the like of the leaves of crops, thereby reducing the photosynthetic efficiency of the soybean leaves and greatly reducing the yield and the quality of the crops; for soybean fruits, the existence of plant diseases and insect pests can cause the phenomena of pod surface necrosis and scab, peculiar and malformed soybean seeds, soft rot and internal dry rot of the soybean seeds and the like of the fruits, thereby directly influencing the quality of the soybeans, reducing the mouthfeel of the soybeans, and reducing the yield of the soybeans.
Therefore, the method has important practical significance and guidance value for identifying the diseases and insect pests and efficiently and accurately identifying the types of the diseases and insect pests, determining the pathogeny and the course of the soybean diseases and insect pests, evaluating the yield and economic loss caused by the diseases and the pests, and taking subsequent control measures aiming at the types of the diseases and the pests. The existing soybean pest and disease identification and identification method usually adopts the Koch formula rule to carry out visual field observation and preliminary diagnosis of symptoms on a plant part with a disease, and then combines laboratory microscopic measurement and analysis of physicochemical parameters of the plant with the disease to determine the type and the cause of the pest and disease, the measurement usually needs professional personnel, the determination process has certain hysteresis, and phenomena such as misdiagnosis or missed diagnosis can also exist in the measurement process, so that the soybean affected by the pest and disease is greatly lost in the best treatment period by mistake.
Disclosure of Invention
Based on the defects, the invention provides a novel method for effectively identifying soybean diseases and insect pests, and solves the problems of hysteresis and inaccuracy in identification and identification of soybean diseases and insect pests in the prior art based on field visual observation, laboratory analysis and the like.
The technology adopted by the invention is as follows: an effective identification method for soybean diseases and insect pests comprises the following steps:
step 1, standardized acquisition of soybean organ images
Acquiring standardized images of soybeans affected by plant diseases and insect pests in different growth periods, and acquiring standardized images of leaf parts, root parts and seed parts of soybean plants affected by the plant diseases and insect pests;
step 2, standardizing the soybean organ images
Performing geometric distortion correction on standardized pictures obtained from soybean organ images in different growth periods, removing geometric deformation brought to the pictures in the shooting process, and selecting uniform sizes for different organ images to perform cutting operation, wherein the standard sizes comprise a blade standardized image, a root standardized image and a seed standardized image of a soybean plant;
step 3, obtaining soybean pest and disease damage identification parameters
The method comprises the following steps of extracting characteristic parameters of soybean organs affected by plant diseases and insect pests in different growth periods, wherein the characteristic parameters of soybean leaves are obtained, the characteristic parameters of soybean root systems are obtained, and the characteristic parameters of soybean seeds are obtained;
step 4, establishing soybean pest and disease damage category database
Establishing a database by using known soybean plant characteristic maps under the influence of different types of soybean diseases and insect pests, extracting and storing various characteristic parameters related in the step (3) under different growth periods of soybean plants under the influence of each disease and pest in the database, endowing different weight to different characteristic parameters according to the difference of the types of the diseases and the difference of the influence of the types of the diseases and pest on the plants, and establishing a distinguishing model of the types of the diseases and pest according to the extraction result of the characteristic parameters;
step 5, identifying and identifying soybean diseases and insect pests
The method comprises the steps of identifying and identifying the actually affected soybean plant diseases and insect pests by using plant characteristic parameters affected by different types of soybean plant diseases and insect pests in different growth periods in a database and discrimination models of the plant disease and pest types in the database, carrying out comparison calculation on characteristic parameter extraction results of the soybean plants to be identified and existing characteristic parameters of the plants under each disease and pest in the database, and finally carrying out comparison calculation and matching on the calculation results of the discrimination models of the characteristic parameters of the soybean plants to be identified and the calculation results of the discrimination models of the plants under each disease and pest, thereby realizing the identification and identification of the types of the soybean plant diseases and insect pests.
The invention also has the following technical characteristics:
1. step 3 as described above: further processing standardized leaf images of soybean plants affected by plant diseases and insect pests in key growth periods of 4 soybeans in flowering, pod bearing, seed swelling and mature periods, extracting corresponding characteristic parameters from the standardized soybean leaf images with the plant disease and insect pest type identification, selecting 20 plants for removing heterogeneity of characteristic parameter extraction results among samples of the leaves of the soybean plants affected by the plant diseases and insect pests to be identified in a specific growth period, randomly selecting a standardized color image of each plant, extracting image matrixes R (x, y), G (x, y) and B (x, y) of three components of red, green and blue of the standardized leaf color image, and converting the standardized leaf color images into gray level images according to a formula F (x, y) R (x, y)/3+ G (x, y)/3+ B (x, y)/3, histogram system for grey scale mapCounting, finding a gray value t1 corresponding to the lowest frequency point between peaks of two gray parts of a green part of the soybean leaf and a color-changing part affected by diseases and insect pests, setting the brightness values of all pixels with the gray values larger than a threshold t1 as 0, and setting the brightness values of more pixels with the gray values smaller than a threshold t1 as 1, so as to convert a gray map into a binary map; according to the number value w1 of the pixel corresponding to the width of the leaf image and the measuring result w1 'cm of the plant corresponding to the leaf image, calculating the actual side length z1 of each pixel as w 1'/w 1 cm, and the actual area of each pixel as z1 2 Square centimeter, according to the actual side length z1 cm of the pixel, and the actual area z1 of the pixel 2 Calculating the mean value of characteristic parameters of the leaves of the plants to be identified in the specific growth period, wherein the mean value comprises the area ratio a1 of the leaves of the plants to be identified, the number a2 of the leaves of the plants to be identified, the area value a3 of the leaves of the plants to be identified and the circumference value a4 of the leaves of the plants to be identified; performing histogram statistics on the gray level image of the plant affected by the plant diseases and insect pests again to further extract the overall morphological characteristics of the leaf affected by the plant diseases and insect pests, finding a gray level t1 ' corresponding to the lowest gray level frequency point between the peak values of the two gray level parts of the soybean leaf and the background color, setting the brightness values of all pixels with the gray levels larger than a threshold value t1 ' as 0, setting the brightness values of all pixels with the gray levels smaller than t1 ' as 1, realizing binarization processing on the whole soybean leaf, and calculating the length-width ratio a5 of the leaf of the plant with the specific growth expectation and the ellipticity ratio a6 of the leaf of the plant with the specific growth expectation according to the number of pixels corresponding to the long axis and the short axis of the leaf binary image;
for the root system of soybean plant affected by plant diseases and insect pests in a certain specific growth period, in order to remove heterogeneity of index parameter extraction results among samples, selecting root system standardized images of 20 plants, extracting image matrixes R (x, y), G (x, y) and B (x, y) of red, green and blue components of a leaf color image, converting the color leaf image into a gray scale image according to a formula F (x, y) which is R (x, y)/3+ G (x, y)/3+ B (x, y)/3, carrying out histogram statistics on the root system gray scale image, and finding out the root system sum of soybean plantsSetting the brightness value of all pixels with the gray scale larger than a threshold value t2 as 0 and the brightness value of all pixels with the gray scale smaller than t2 as 1, converting the root system gray scale image into a binary image, calculating the actual side length z2 of each pixel as w2 '/w 2 cm according to the pixel number value w2 corresponding to the root system image width and the measurement result w 2' cm of the plant corresponding to the root system image, and setting the actual area of each pixel as z2 2 Square centimeter, according to the actual side length z2 centimeter of the pixel and the actual area z2 of the pixel 2 Square centimeter and the number of pixels corresponding to different parameter extraction results of plant roots affected by diseases and insect pests, calculating a root area value b1 of the plant with the expected identification of the specific growth expectation, a root nodule number value b2 of the plant root system of the plant with the diseases and insect pests to be identified, calculating a mean value of root texture characteristic parameters of the plant with the diseases and pests to be identified in the specific growth period according to a gray scale map, wherein the specific calculation process of the root texture characteristic parameters comprises the steps of calculating a root contrast texture characteristic value b3 of the plant with the diseases and pests to be identified according to a formula 1, calculating a root consistency texture characteristic value b4 of the plant with the diseases and pests to be identified according to a formula 2, calculating an entropy texture characteristic value b5 of the plant root system of the plant with the diseases and pests to be identified according to a formula 3, and calculating an energy texture characteristic value b6 of the plant root system of the plant with the diseases and pests to be identified according to a formula 4,
p (i, j) represents the value of a gray level co-occurrence matrix calculated by a root system gray level image at the position of the ith row and the jth column, represents the probability of simultaneous occurrence of a pixel with a gray level i and a pixel with a gray level j in the root system gray level image in a fixed direction and pixel interval, n is the difference of the two gray levels i and j in the root system gray level image, and Ng represents the stage number of the gray level co-occurrence matrix extracted by the root system gray level image;
for seeds of soybean plants affected by pests and diseases in a certain specific growth period, in order to remove heterogeneity of index parameter extraction results among samples, 20 plants are selected, image matrixes R (x, y), G (x, y) and B (x, y) of three components of red, green and blue of a color seed image of each plant seed are extracted, the color seed image is converted into a gray scale map according to a formula F (x, y) ═ R (x, y)/3+ G (x, y)/3+ B (x, y)/3 for the three components, histogram statistics is carried out on the gray scale map to find a gray scale value t3 corresponding to a frequency lowest point between peaks of two gray scale parts of a soybean seed part and a background part of a test bench, all pixel brightness values of which the gray scale is greater than a threshold t3 are set as 1, and most pixel brightness values of which the gray scale is smaller than the threshold t3 are set as 0, so as to convert the gray scale map into a binary map, calculating the actual side length z3 of each pixel to be w3 '/w 3 cm according to the pixel number value w3 corresponding to the seed image width influenced by plant diseases and insect pests and the measurement result w 3' cm of the plant corresponding to the root system image, wherein the actual area of each pixel is z3 2 Square centimeter, according to the actual side length z3 centimeter of the pixel and the actual area z3 of the pixel 2 The method comprises the steps of square centimeter and pixel numbers corresponding to different parameter extraction results of plant seeds affected by diseases and insect pests, calculating the mean value of characteristic parameters of the plant seeds affected by the diseases and insect pests to be identified in a specific growth period, wherein the mean value comprises the numerical value c1 of the plant seeds affected by the diseases and insect pests to be identified, the area value c2 of the plant seeds affected by the diseases and insect pests to be identified, the circumferential length value c3 of the plant seeds affected by the diseases and insect pests to be identified, the length-width ratio c4 of the plant seeds affected by the diseases and insect pests to be identified, the curvature value c5 of the plant seeds affected by the diseases and insect pests to be identified and the ellipse value c6 of the plant seeds affected by the diseases and insect pests to be identified.
2. Step 4 as described above: establishing characteristic parameter database by using soybean plant images affected by all known plant diseases and insect pests in different growth periods of flowering period, pod bearing period, grain swelling period and mature period, and establishing characteristic parameter database for specific growth periodExtracting and establishing a library of soybean leaf characteristic parameters under the influence of internal known plant diseases and insect pests, converting a color leaf image under the influence of the known plant diseases and insect pests in a specific growth period into a gray map by combining the method in the step 3, performing histogram statistics on the gray map by using the method in the step 3 to find a gray value T1 corresponding to a frequency lowest point between peaks of two gray parts of a green part of a leaf and a color-changing part under the influence of the plant diseases and insect pests, setting the brightness value of all pixels with the gray value larger than a threshold T1 as 0, setting the brightness value of more pixels with the gray value smaller than the threshold T1 as 1, converting the gray map into a binary map, calculating the actual side length Z1 of each pixel to be W1 '/W1 cm according to the pixel number W1 corresponding to the width of the leaf image and the measurement result W1' cm of a plant corresponding to the leaf image, and calculating the actual area of each pixel to be Z1 2 Square centimeter, according to the actual side length Z1 centimeter of the pixel and the actual area Z1 of the pixel 2 Square centimeter and the number of pixels corresponding to the extraction result of different parameters of each spot, calculating the mean value of the characteristic parameters of the plant leaf of the known pest type in the specific growth period, including the ratio A1 of the area of the leaf spot of the known pest plant, the number A2 of the leaf spot of the known pest plant, the value A3 of the leaf spot area of the known pest plant, and the value A4 of the circumference of the leaf spot of the known pest plant, in order to further extract the overall morphological characteristics of the leaf affected by the known type of pest, performing histogram statistics on the gray map of the plant affected by the known pest, finding the gray value T1 ' corresponding to the lowest gray level frequency point between the two peak values of the gray level parts of the soybean leaf and the background color, setting the brightness values of all the pixels with the gray level greater than the threshold value T1 ' to be 0, setting the brightness values of all the pixels with the gray level less than T1 ' to be 1, and realizing the binarization processing of the whole soybean leaf, extracting the number of pixels corresponding to the long axis and the short axis of the leaf according to the binary leaf map, and calculating the ratio A5 of the length and the width of the leaf spot of the known plant diseases and insect pests in the specific growth period and the value A6 of the ellipse spot of the known plant diseases and insect pests;
extracting and establishing a database of soybean root characteristic parameters under the influence of known plant diseases and insect pests in a specific growth period, and converting color root images under the influence of known plant diseases and insect pests in the specific growth period by the method of step 3Performing histogram statistics on the gray scale image by using the method in the step 3 to find a gray scale value T2 corresponding to a gray scale frequency lowest point between two gray scale part peaks of a soybean root system and a background color, setting the brightness value of all pixels with the gray scale larger than a threshold value T2 as 0, setting the brightness value of all pixels with the gray scale smaller than T2 as 1, converting the root system gray scale image into a binary image, calculating the actual side length Z2 of each pixel as W2 '/W2 cm according to the pixel number value W2 corresponding to the width of the root system image and the measurement result W2' cm of a plant corresponding to the root system image, and setting the actual area of each pixel as Z2 2 Square centimeter, according to the actual side length Z2 centimeter of the pixel and the actual area Z2 of the pixel 2 Square centimeter and pixel numbers corresponding to different extraction results of parameters of plant roots affected by diseases and insect pests, calculating the mean value of all characteristic parameters of the plant roots affected by the known diseases and pests, including the area value B1 of the plant roots of the known diseases and pests and the root nodule number B2 of the plant roots of the known diseases and pests, calculating the contrast texture characteristic value B3 of the plant roots of the known diseases and pests according to the formula 1 in the third step, calculating the consistency texture characteristic value B4 of the plant roots of the known diseases and pests according to the formula 2, calculating the entropy texture characteristic value B5 of the plant roots of the known diseases and pests according to the formula 3, and calculating the energy texture characteristic value B6 of the plant roots of the known diseases and pests according to the formula 4;
for the seeds of soybean plants affected by known plant diseases and insect pests in a certain specific growth period, converting a color seed image affected by the known plant diseases and insect pests in the specific growth period into a gray map by combining the method in the step 3, performing histogram statistics on the gray map to find a gray value T3 corresponding to the lowest frequency point between the peak values of two gray parts of a soybean seed part and a background part of a test bench, setting the brightness value of all pixels with the gray value larger than a threshold value T3 as 1, setting the brightness value of most pixels with the gray value smaller than the threshold value T3 as 0, converting the gray map of the seeds into a binary map, calculating the actual side length Z3 ═ W3 '/W3 cm of each pixel according to the pixel number W3 corresponding to the width of the seed image affected by the plant diseases and insect pests and the measurement result W3' cm of the plants corresponding to the root system image, and setting the actual area of each pixel as Z3 2 Square centimeter, according to the actual side length Z3 centimeter of the pixel and the actual surface of the pixelProduct Z3 2 Square centimeter, and the pixel number corresponding to the extraction result of different parameters of plant seeds affected by plant diseases and insect pests, calculating the mean value of characteristic parameters of the known plant seeds affected by the plant diseases and insect pests in the specific growth period, wherein the mean value comprises the numerical value C1 of the known plant seeds affected by the plant diseases and insect pests, the area value C2 of the known plant seeds affected by the plant diseases and insect pests, the girth value C3 of the known plant seeds affected by the plant diseases and insect pests, the length-width ratio C4 of the known plant seeds affected by the plant diseases and insect pests, the curvature value C5 of the known plant seeds, the ellipse curvature value C6 of the known plant seeds affected by the plant diseases and insect pests, and determining the weight K of the characteristic parameters of each known soybean plant type of plant diseases and insect pests according to the difference of the different types of plant diseases and insect pests on the organs of the soybean plants 1 Determining the weight K of the root characteristic parameter of each known soybean pest type 2 Determining the weight K of each known soybean pest type seed characteristic parameter 3 While satisfying the normalized characteristic of the weight parameter, i.e. K 1 +K 2 +K 3 1, establishing a pest and disease distinguishing model according to the extracted different organ image characteristic parameter extraction results of the soybean pest and disease types to be identified and the different organ image characteristic parameter extraction results of the known pest and disease types in the database, wherein the distinguishing model is shown as a formula 5:
4. step 5 as described above: identifying and identifying the types of the affected soybean plant diseases and insect pests by using plant characteristic parameters affected by different types of known soybean plant diseases and insect pests in different growth periods in the database in the step 4 and known plant disease and pest type distinguishing models in the database, and extracting results a1, a2, a3, a4, a5 and a6 of the characteristic parameters of leaves of a certain plant to be identified; extracting results b1, b2, b3, b4, b5 and b6 of the characteristic parameters of the roots of plants suffering from the plant diseases and insect pests to be identified; and (3) extracting results c1, c2, c3, c4, c5 and c6 of seed characteristic parameters of plants to be identified affected by the plant diseases and insect pests, bringing the characteristic parameter extraction result of each soybean to be identified affected by the plant diseases and insect pests and the characteristic parameter extraction result of each known soybean affected by the plant diseases and insect pests in the growing period into the judgment model in the step (4), calculating the sizes of P values of model calculation results of the types of the plant diseases and insect pests to be identified and all known plant disease and insect pest types according to the judgment model, and proving that the closer the type of the plant diseases and insect pests to be identified is to the known plant disease and insect pest type corresponding to the P value calculation result, so that the identification and identification of the types of the plant diseases and insect pests of the soybeans are realized.
The invention has the following beneficial effects and advantages: compared with the laboratory analysis and identification, the method has the characteristics of low identification cost, low manufacturing cost, small loss, small damage to plants, real-time performance and quasi-real-time performance, and the identification personnel can quickly acquire the soybean plant characteristic parameters of the plant diseases and insect pests to be identified only through images and realize the identification of the plant diseases and insect pests according to the extraction result. In addition, the requirement on the professional knowledge of the identifier is low, the identifier can identify and identify the soybean disease and insect pest types to be identified by utilizing a large amount of data of various known disease and insect pest types in the database only through simple training by utilizing the method, and the identifier does not need to comprehensively master the disease and insect pest knowledge in the whole identification process. In addition, compared with the traditional method for identifying the soybean plant diseases and insect pests by visual inspection in the field, the method disclosed by the invention is high in identification speed, and because the automation degree of the extraction process of the plant parameters to be identified is high, the method can simultaneously process a large number of soybean plant images of the plant diseases and insect pests to be identified in batches by a programming means and extract the soybean plant parameters of the plant diseases and insect pests to be identified, so that the quick and efficient batch identification of the plant diseases and insect pests of the soybean plants is realized.
Drawings
FIG. 1 is a gray scale image of plant leaves affected by plant diseases and insect pests;
FIG. 2 is a binary image of the interior of a plant leaf affected by plant diseases and insect pests;
FIG. 3 is a binary image of the whole plant leaves affected by plant diseases and insect pests;
FIG. 4 is a gray scale view of the plant root system affected by plant diseases and insect pests;
FIG. 5 is a binary map of plant root systems affected by pests;
FIG. 6 is a gray scale diagram of plant seeds affected by plant diseases and insect pests;
FIG. 7 is a binary image of seeds of plants affected by plant diseases and insect pests;
Detailed Description
The invention is further illustrated by way of example in the accompanying drawings of the specification:
example 1
Step 1: obtaining typical soybean plant organ standardized images of soybeans affected by plant diseases and insect pests in different growth periods such as flowering period, pod bearing period, grain swelling period, mature period and the like,
for soybean leaf images affected by diseases and insect pests for a certain lifetime, a digital camera is installed on a fixed support, the leaves affected by the diseases and insect pests are placed on a test bed, the position of the support is adjusted to enable a camera lens to be perpendicular to the test bed, the camera lens is 20cm away from the soybean leaves, a photographing environment is determined and adjusted by a photometer, a whole leaf affected by the diseases and insect pests is photographed to obtain a standardized image of the leaf, and a chessboard grid calibration plate image is photographed;
for soybean root system images affected by diseases and insect pests for a certain lifetime, a digital camera is mounted on a fixed support, the position of the camera is adjusted, a camera lens is perpendicular to the ground surface, soybean plants are flatly paved on the ground surface, the height of the camera lens from the ground surface is ensured to be 1m, a photographing environment is determined and adjusted by a photometer, and the large root system is photographed for acquiring soybean root system standardized images affected by the diseases and insect pests and photographing chessboard grid calibration plate images;
for soybean seed grain images affected by diseases and insect pests for a certain lifetime, pods affected by the diseases and insect pests are placed on a test bed, the position of a support is adjusted to enable a camera lens to be perpendicular to the test bed, then soybean seed grains of the whole plant are taken out and placed on the test bed in a dispersing mode, the position of the support is adjusted to enable the camera lens to be perpendicular to the test bed, the height of the camera lens is 20cm away from the soybean seed grains, a photographing environment is determined and adjusted by a photometer, standardized images of soybean grains affected by the diseases and insect pests are photographed, and chessboard grid calibration plate images are photographed.
Step 2: in the step 1, the extraction results of various types of soybean organ images affected by diseases and insect pests are subjected to standardized processing, and for the various types of soybean organ images, because the standardized images are identical to the corresponding calibration board image shooting conditions, the standardized images of the various types of soybean organs are identical to the geometric distortion of the corresponding black and white chessboard calibration board images.
And step 3: and further processing blade standardized images of soybean plants affected by diseases and insect pests in 4 key soybean growth periods of flowering stage, pod bearing stage, grain swelling stage and mature stage, and extracting corresponding characteristic parameters from the soybean blade standardized images for identifying the types of the diseases and insect pests.
For the leaves of soybean plants affected by the diseases and insect pests to be identified in a certain specific growth period, 20 plants are selected in order to remove the heterogeneity of characteristic parameter extraction results among samples, and each plant randomly selects a standardized color image of the leaf. The image matrixes R (x, y), G (x, y), B (x, y) of the three components of red, green and blue of the blade normalized color image are extracted, and then the normalized color image of the blade is converted into a gray map for the three components according to the formula F (x, y) ═ R (x, y)/3+ G (x, y)/3+ B (x, y)/3, and the gray map extraction result is shown in fig. 1. Histogram statistics is carried out on the gray level image, a gray level t1 corresponding to the lowest frequency point between peaks of two gray level parts of a green part of the soybean leaf and a color change part affected by plant diseases and insect pests is found, brightness values of all pixels with the gray levels larger than a threshold t1 are set to be 0, brightness values of more pixels with the gray levels smaller than a threshold t1 are set to be 1, and the gray level image is converted into a binary image as shown in fig. 2. Calculating the actual side length z1 of each pixel to be w1 '/w 1 cm according to the pixel number value w1 corresponding to the width of the leaf image and the measurement result w1 ' cm of the plant corresponding to the leaf image, wherein the actual side length of each pixel is w1 '/w 1 cmThe actual area is z1 2 Square centimeter, according to the actual side length z1 centimeter of the pixel and the actual area z1 of the pixel 2 The method comprises the steps of calculating the mean value of characteristic parameters of the leaves of the plant to be identified by the diseases and insect pests in a specific growth period by square centimeters and the pixel numbers corresponding to different parameter extraction results of each spot, wherein the mean value comprises the area ratio a1 of the leaves of the plant to be identified by the diseases and insect pests, the number a2 of the leaves of the plant to be identified by the diseases and insect pests, the area value a3 of the leaves of the plant to be identified by the diseases and insect pests, and the circumference value a4 of the leaves of the plant to be identified by the diseases and insect pests.
In order to further extract the overall morphological characteristics of the leaves affected by the plant diseases and insect pests, histogram statistics is carried out on a gray map of the plant affected by the plant diseases and insect pests again, a gray value t1 ' corresponding to the lowest gray level frequency point between the peaks of the two gray level parts of the soybean leaves and the background color is found, the brightness value of all pixels with the gray levels larger than a threshold value t1 ' is set to 0, the brightness value of all pixels with the gray levels smaller than t1 ' is set to 1, and the binarization processing of the whole soybean leaves is realized, for example, fig. 3 shows the binarization processing result of the whole soybean leaves, the length-width ratio a5 of the leaves of the plant with the specific growth expectation and the plant leaf ellipticity a6 of the plant with the specific growth expectation and identification are extracted according to the binary map of the leaves of the soybean.
For the root system of soybean plants affected by plant diseases and insect pests in a certain growth period, in order to remove the heterogeneity of index parameter extraction results among samples, root system standardized images of 20 plants are selected, image matrixes R (x, y), G (x, y), B (x, y) of red, green and blue components of a leaf color image are extracted, and the color leaf image is converted into a gray scale image according to a formula F (x, y) ═ R (x, y)/3+ G (x, y)/3+ B (x, y)/3 for the three components, as shown in fig. 4.
Histogram statistics is carried out on the root system gray image, a gray value t2 corresponding to the lowest gray level frequency point between the two peak values of the gray level part of the soybean root system and the background color is found, the brightness value of all pixels with the gray level larger than a threshold t2 is set to be 0, the brightness value of all pixels with the gray level smaller than t2 is set to be 1, and the root system gray image is converted into a binary image, as shown in fig. 5.
According to root system image width correspondenceThe number value w2 of the pixels and the measurement result w2 'cm of the plants corresponding to the root system images are calculated, the actual side length z2 of each pixel is w 2'/w 2 cm, and the actual area of each pixel is z2 2 Square centimeter, according to the actual side length z2 cm of the pixel, and the actual area z2 of the pixel 2 The method comprises the steps of calculating the area value b1 of the root system of the plant suffering from the diseases and insect pests expected to be identified in the specific growth period, calculating the root system root nodule number value b2 of the plant suffering from the diseases and insect pests to be identified, calculating the mean value of the root system texture characteristic parameters of the plant suffering from the diseases and insect pests to be identified in the specific growth period according to a gray scale map, wherein the specific calculation process of the root system texture characteristic parameters is as follows, calculating the root system contrast texture characteristic value b3 of the plant suffering from the diseases and insect pests to be identified according to a formula 1, calculating the root system consistency texture characteristic value b4 of the plant suffering from the diseases and insect pests to be identified according to a formula 2, calculating the root system entropy texture characteristic value b5 of the plant suffering from the diseases and insect pests to be identified according to a formula 3, and calculating the root system energy texture characteristic value b6 of the plant suffering from the diseases and insect pests to be identified according to a formula 4.
P (i, j) represents the value of a gray level co-occurrence matrix calculated by the root system gray level image at the position of the ith row and the jth column, represents the probability of the simultaneous occurrence of a pixel with the gray level i and a pixel with the gray level j in the root system gray level image in a fixed direction and pixel interval, n is the difference of the two gray levels i and j in the root system gray level image, and Ng represents the stage number of the gray level co-occurrence matrix extracted by the root system gray level image.
For the seeds of soybean plants affected by plant diseases and insect pests in a certain specific growth period, in order to remove the heterogeneity of index parameter extraction results among samples, 20 plants are selected, image matrixes R (x, y), G (x, y), B (x, y) of red, green and blue components of a color seed image of each plant seed are extracted, the color seed image is converted into a gray scale image according to a formula F (x, y) ═ R (x, y)/3+ G (x, y)/3+ B (x, y)/3 for the three components, as shown in fig. 6,
histogram statistics is carried out on the gray level image to find out the gray level t3 corresponding to the frequency lowest point between the peak values of the two gray level parts of the soybean seed part and the background part of the test bed, the brightness values of all pixels with the gray levels larger than a threshold t3 are set to be 1, the brightness values of more pixels with the gray levels smaller than a threshold t3 are set to be 0, and the gray level image of the seeds is converted into a binary image, as shown in fig. 7.
Calculating the actual side length z3 of each pixel to be w3 '/w 3 cm according to the pixel number value w3 corresponding to the seed image width influenced by plant diseases and insect pests and the measurement result w 3' cm of the plant corresponding to the root system image, wherein the actual area of each pixel is z3 2 Square centimeter, according to the actual side length z3 centimeter of the pixel and the actual area z3 of the pixel 2 Square centimeter to and the pixel number that the different parameters of plant seed grain that are influenced by plant diseases and insect pests draw the result and correspond, calculate the mean value of waiting to appraise in this specific growth period and receive plant seed grain characteristic parameter, including waiting to appraise plant seed grain number of a plant disease and insect pests value c1, wait to appraise plant disease and insect pests seed grain area value c2, wait to appraise plant disease and insect pests seed grain girth value c3, wait to appraise plant disease and insect pests seed grain length-width ratio c4, wait to appraise plant disease and insect pests seed grain curvature value c5, wait to appraise plant disease and insect pests seed grain ellipse ratio value c 6.
And 4, step 4: establishing a characteristic parameter database by using soybean plant images affected by all known plant diseases and insect pests in different growth periods of the flowering period, the pod bearing period, the grain swelling period and the mature period, extracting characteristic parameters of soybean leaves affected by known plant diseases and insect pests in a specific growth period, establishing the database, and converting the color leaf images affected by the known plant diseases and insect pests in the specific growth period by combining the method and the formula in the step 3Performing histogram statistics on the gray scale image by using the method in the step 3 to find a gray scale value T1 corresponding to a frequency lowest point between peaks of two gray scale parts of a green part of the leaf and a color change part affected by diseases and insect pests, setting the brightness value of all pixels with gray scales larger than a threshold T1 as 0, setting the brightness value of a plurality of pixels with gray scales smaller than a threshold T1 as 1, converting the gray scale image into a binary image, calculating the actual side length Z1 of each pixel as W1 '/W1 cm according to the pixel number value W1 corresponding to the width of the leaf image and the measurement result W1' cm of a plant corresponding to the leaf image, and calculating the actual area of each pixel as Z1 2 Square centimeter, according to the actual side length Z1 cm of the pixel and the actual area Z1 of the pixel 2 Square centimeter and the number of pixels corresponding to the extraction result of different parameters of each spot, calculating the mean value of the characteristic parameters of the plant leaf of the known pest type in the specific growth period, including the ratio A1 of the area of the leaf spot of the known pest plant, the number A2 of the leaf spot of the known pest plant, the value A3 of the leaf spot area of the known pest plant, and the value A4 of the circumference of the leaf spot of the known pest plant, in order to further extract the overall morphological characteristics of the leaf affected by the known type of pest, performing histogram statistics on the gray map of the plant affected by the known pest, finding the gray value T1 ' corresponding to the lowest gray level frequency point between the two peak values of the gray level parts of the soybean leaf and the background color, setting the brightness values of all the pixels with the gray level greater than the threshold value T1 ' to be 0, setting the brightness values of all the pixels with the gray level less than T1 ' to be 1, and realizing the binarization processing of the whole soybean leaf, and extracting the pixel number corresponding to the long axis and the pixel number corresponding to the short axis of the leaf according to the binary leaf map, and calculating the ratio A5 of the length and the width of the leaf spot of the known plant diseases and insect pests in the specific growth period and the value A6 of the ellipse spot of the known plant diseases and insect pests.
Extracting and establishing a database of soybean root characteristic parameters under the influence of known plant diseases and insect pests in a certain specific growth period by combining the method and the formula in the step 3, converting the color root image under the influence of the known plant diseases and insect pests in the specific growth period into a gray level image, performing histogram statistics on the gray level image by using the method in the step 3, and finding out the lowest gray level frequency between the peak values of the two gray level parts of the soybean root and the background colorSetting the brightness values of all pixels with the gray scale larger than a threshold value T2 as 0 and the brightness values of all pixels with the gray scale smaller than T2 as 1, converting the root system gray scale image into a binary image, calculating the actual side length Z2 of each pixel as W2 '/W2 cm according to the pixel number value W2 corresponding to the width of the root system image and the measurement result W2' cm of the plant corresponding to the root system image, and setting the actual area of each pixel as Z2 2 Square centimeter, according to the actual side length Z2 centimeter of the pixel and the actual area Z2 of the pixel 2 The method comprises the steps of calculating the mean value of all characteristic parameters of the plant root system affected by known plant diseases and insect pests, wherein the mean value comprises a known plant disease and insect pests root system area value B1 and a known plant disease and insect pests root system root nodule number value B2, calculating a known plant disease and insect pests root system contrast texture characteristic value B3 according to a formula 1 in the third step, calculating a known plant disease and insect pests root system consistency texture characteristic value B4 according to a formula 2 in the third step, calculating a known plant disease and insect pests root system entropy texture characteristic value B5 according to a formula 3 in the third step, and calculating a known plant disease and insect pests root system energy texture characteristic value B6 according to a formula 4 in the third step.
For the seeds of soybean plants affected by known plant diseases and insect pests in a certain specific growth period, converting a color seed image affected by the known plant diseases and insect pests in the specific growth period into a gray map by combining the method and the formula in the step 3, carrying out histogram statistics on the gray map to find a gray value T3 corresponding to a frequency lowest point between peaks of two gray parts of a soybean seed part and a background part of a test bench, setting the brightness value of all pixels with the gray value larger than a threshold value T3 as 1, setting the brightness values of more pixels with the gray value smaller than the threshold value T3 as 0, converting the gray map of the seeds into a binary map, calculating the actual side length Z3 of each pixel to be W3 '/W3 cm according to the pixel value W3 corresponding to the width of the seed image affected by the plant diseases and the measurement result W3' cm of plants corresponding to root images, and setting the actual area of each pixel to be Z3 2 Square centimeter, according to the actual side length Z3 centimeter of the pixel and the actual area Z3 of the pixel 2 Square centimeter and corresponding to different parameter extraction results of plant grains affected by plant diseases and insect pestsCalculating the mean value of characteristic parameters of plant seeds which are known to be affected by diseases and insect pests in the specific growth period, wherein the mean value comprises a seed number value C1 of a plant with known diseases and insect pests, an area value C2 of the seed of the plant with known diseases and insect pests, a girth value C3 of the seed of the plant with known diseases and insect pests, a length-width ratio C4 of the seed of the plant with known diseases and insect pests, a seed curvature value C5 of the plant with known diseases and insect pests and an ellipse curvature value C6 of the seed of the plant with known diseases and insect pests. Determining the weight K of the characteristic parameters of the leaves of each known soybean pest type according to the difference of the influence of different types of pests on the organs of the soybean plants 1 Determining the weight K of the root characteristic parameter of each known soybean pest type 2 Determining the weight K of each known soybean pest type seed characteristic parameter 3 While satisfying the normalized characteristic of the weight parameter, i.e. K 1 +K 2 +K 3 1. And establishing a pest and disease distinguishing model according to the extracted different organ image characteristic parameter extraction results of the types of soybean pests to be identified and the different organ image characteristic parameter extraction results of the types of known pests in the database, wherein the distinguishing model is shown as a formula 5.
And 5: identifying and identifying the types of the affected soybean plant diseases and insect pests by using plant characteristic parameters affected by different types of known soybean plant diseases and insect pests in different growth periods in the database in the step 4 and known plant disease and pest type distinguishing models in the database, and extracting results a1, a2, a3, a4, a5 and a6 of the characteristic parameters of leaves of a certain plant to be identified; extracting results b1, b2, b3, b4, b5 and b6 of the characteristic parameters of the roots of plants suffering from the plant diseases and insect pests to be identified; and extracting characteristic parameters of seeds of plants suffering from the plant diseases and insect pests to be identified to obtain results c1, c2, c3, c4, c5 and c 6. And (4) bringing the characteristic parameter extraction result of each soybean affected by the diseases and the insect pests to be identified and the characteristic parameter extraction result of each known soybean affected by the diseases and the insect pests in the growth period into the judgment model in the step (4), calculating the sizes of model calculation results P of the types of the diseases and the insect pests to be identified and all the known types of the diseases and the insect pests according to the judgment model, wherein the smaller the P value calculation result is, the closer the type of the diseases and the insect pests to be identified is to the known type of the diseases and the insect pests corresponding to the P value calculation result is proved, and therefore the quick, accurate and effective identification and identification work of the types of the diseases and the insect pests of the soybeans is realized.
Claims (3)
1. An effective identification method for soybean diseases and insect pests is characterized by comprising the following steps:
step 1, standardized acquisition of soybean organ images
Acquiring standardized images of soybeans affected by plant diseases and insect pests in different growth periods, and acquiring standardized images of leaf parts, root parts and seed parts of soybean plants affected by the plant diseases and insect pests;
step 2, standardizing the soybean organ images
Performing geometric distortion correction on standardized pictures obtained from soybean organ images in different growth periods, removing geometric deformation brought to the pictures in the shooting process, and selecting uniform sizes for different organ images to perform cutting operation, wherein the standard sizes comprise a blade standardized image, a root standardized image and a seed standardized image of a soybean plant;
step 3, obtaining soybean pest and disease damage identification parameters
The method comprises the following steps of extracting characteristic parameters of soybean organs affected by plant diseases and insect pests in different growth periods, wherein the characteristic parameters of soybean leaves, the characteristic parameters of soybean root systems and soybean seeds are obtained, and the specific method comprises the following steps: further processing blade standardized images obtained by soybean plants affected by diseases and insect pests in 4 key soybean growth periods of flowering phase, pod bearing phase, seed swelling phase and mature phase, extracting corresponding characteristic parameters aiming at each blade standardized image for identifying the types of the diseases and the pests, selecting 20 plants aiming at the blades of the soybean plants affected by the diseases and the pests to be identified in a certain specific growth period and removing the heterogeneity of the extraction results of the characteristic parameters between samples, randomly selecting a standardized color image of each blade for each plant, and extracting red, green and blue of the standardized color image of the blades,The method comprises the steps of obtaining an image matrix R (x, y), G (x, y) and B (x, y) of three components of green and blue, converting a standardized color image of a leaf into a gray map according to a formula F (x, y) ═ R (x, y)/3+ G (x, y)/3+ B (x, y)/3, carrying out histogram statistics on the gray map, finding a gray value t1 corresponding to a lowest frequency point between peaks of two gray parts of a green part of the soybean leaf and a color-changing part affected by diseases and insect pests, setting the brightness values of all pixels with the gray levels larger than a threshold t1 as 0, setting the brightness values of most pixels with the gray levels smaller than a threshold t1 as 1, and converting the gray map into a binary map; calculating the actual side length z1 of each pixel (w 1 '/w 1 cm) according to the pixel number value w1 corresponding to the width of the leaf image and the measurement result w 1' cm of the plant corresponding to the leaf image, wherein the actual area of each pixel is z1 2 Square centimeter, according to the actual side length z1 centimeter of the pixel and the actual area z1 of the pixel 2 Calculating the mean value of characteristic parameters of the leaf blade of the plant to be identified with the pest in the specific growth period, wherein the mean value comprises the area ratio a1 of the leaf blade of the plant to be identified with the pest, the number a2 of the leaf blade spot of the plant to be identified with the pest, the area value a3 of the leaf blade of the plant to be identified with the pest and the circumference value a4 of the leaf spot of the plant to be identified with the pest; performing histogram statistics on the gray level image of the plant affected by the plant diseases and insect pests again to further extract the overall morphological characteristics of the leaf affected by the plant diseases and insect pests, finding a gray level t1 ' corresponding to the lowest gray level frequency point between the peak values of the two gray level parts of the soybean leaf and the background color, setting the brightness values of all pixels with the gray levels larger than a threshold value t1 ' as 0, setting the brightness values of all pixels with the gray levels smaller than t1 ' as 1, realizing binarization processing on the whole soybean leaf, and calculating the length-width ratio a5 of the leaf of the plant with the specific growth expectation and the ellipticity ratio a6 of the leaf of the plant with the specific growth expectation according to the number of pixels corresponding to the long axis and the short axis of the leaf binary image;
for the root system of soybean plants affected by diseases and insect pests in a certain specific growth period, in order to remove the heterogeneity of the extraction result of the index parameters among samples, the root system standardized images of 20 plants are selected, and the red, green and blue three components of the color image of the leaves are extractedAn image matrix of quantities R (x, y), G (x, y), B (x, y), for which the color leaf image is converted into a grayscale image according to the formula F (x, y) -R (x, y)/3+ G (x, y)/3+ B (x, y)/3, histogram statistics is carried out on the root system gray image, a gray value t2 corresponding to the lowest gray level frequency point between the two peak values of the gray level frequency of the soybean root system and the background color is found, the brightness value of all pixels with the gray level larger than a threshold value t2 is set as 0, the brightness value of all pixels with the gray level smaller than t2 is set as 1, the root system gray image is converted into a binary image, according to the pixel number value w2 corresponding to the width of the root system image and the measurement result w2 'cm of the plant corresponding to the root system image, the actual side length z2 of each pixel is calculated to be w 2'/w 2 cm, and the actual area of each pixel is z 2. 2 Square centimeter, according to the actual side length z2 centimeter of the pixel and the actual area z2 of the pixel 2 Square centimeter and the number of pixels corresponding to different parameter extraction results of plant roots affected by diseases and insect pests, calculating a root area value b1 of the plant with the expected identification of the specific growth expectation, a root nodule number value b2 of the plant root system of the plant with the diseases and insect pests to be identified, calculating a mean value of root texture characteristic parameters of the plant with the diseases and pests to be identified in the specific growth period according to a gray scale map, wherein the specific calculation process of the root texture characteristic parameters comprises the steps of calculating a root contrast texture characteristic value b3 of the plant with the diseases and pests to be identified according to a formula 1, calculating a root consistency texture characteristic value b4 of the plant with the diseases and pests to be identified according to a formula 2, calculating an entropy texture characteristic value b5 of the plant root system of the plant with the diseases and pests to be identified according to a formula 3, and calculating an energy texture characteristic value b6 of the plant root system of the plant with the diseases and pests to be identified according to a formula 4,
p (i, j) represents the value of a gray level co-occurrence matrix calculated by a root system gray level image at the position of the ith row and the jth column, represents the probability of simultaneous occurrence of a pixel with a gray level i and a pixel with a gray level j in the root system gray level image in a fixed direction and pixel interval, n is the difference of the two gray levels i and j in the root system gray level image, and Ng represents the stage number of the gray level co-occurrence matrix extracted by the root system gray level image;
for seeds of soybean plants affected by pests and diseases in a certain specific growth period, in order to remove heterogeneity of index parameter extraction results among samples, 20 plants are selected, an image matrix R (x, y), G (x, y) and B (x, y) of three components of red, green and blue of a color seed image of each plant seed is extracted, the color seed image is converted into a gray scale map according to a formula F (x, y) which is R (x, y)/3+ G (x, y)/3+ B (x, y)/3, the gray scale map is subjected to histogram statistics to find a gray scale value t3 corresponding to a frequency lowest point between peaks of two gray scale parts of a soybean seed part and a background part of a test table, all pixel brightness values of which are greater than a threshold value t3 are set as 1, most pixel brightness values of which are less than the threshold value t3 are set as 0, and the gray scale map is converted into a binary map, calculating the actual side length z3 of each pixel to be w3 '/w 3 cm according to the pixel number value w3 corresponding to the seed image width influenced by plant diseases and insect pests and the measurement result w 3' cm of the plant corresponding to the root system image, wherein the actual area of each pixel is z3 2 Square centimeter, according to the actual side length z3 centimeter of the pixel and the actual area z3 of the pixel 2 Square centimeter to and the corresponding pixel number of the different parameter extraction results of plant seed grain that is influenced by plant diseases and insect pests, calculate the mean value of waiting to appraise the plant seed grain characteristic parameter of suffering from diseases and insect pests in this specific growth period, including waiting to appraise plant seed grain number value c1, wait to appraise plant seed grain area value c2, wait to appraise plant seed grain girth value c3, wait to appraise plant seed grain length and width ratio c4, wait to appraise plant diseases and insect pests plant seed grain curvature value c5, wait to appraise the plant diseases and insect pestsEllipse rate value c6 of seed of insect plant
Step 4, establishing soybean disease and pest species database
Establishing a database by utilizing the known soybean plant characteristic maps under the influence of different types of soybean plant diseases and insect pests, extracting and storing various characteristic parameters related in the step (3) in the database under different growth periods of the soybean plant under the influence of each type of the plant diseases and insect pests, giving different weights to the different characteristic parameters according to the difference of the types of the plant diseases and the difference of the influence of the plant diseases and insect pests on the different characteristic parameters, and establishing a distinguishing model of the types of the plant diseases and insect pests according to the extraction result of the characteristic parameters;
step 5, identifying and identifying soybean diseases and insect pests
The method comprises the steps of identifying and identifying the actually affected soybean plant diseases and insect pests by using plant characteristic parameters affected by different types of soybean plant diseases and insect pests in different growth periods in a database and discrimination models of the plant disease and pest types in the database, carrying out comparison calculation on characteristic parameter extraction results of the soybean plants to be identified and existing characteristic parameters of the plants under each disease and pest in the database, and finally carrying out comparison calculation and matching on the calculation results of the discrimination models of the characteristic parameters of the soybean plants to be identified and the calculation results of the discrimination models of the plants under each disease and pest, thereby realizing the identification and identification of the types of the soybean plant diseases and insect pests.
2. The method for effectively identifying soybean pests and diseases according to claim 1, which is characterized in that:
and 4, step 4: establishing a characteristic parameter database by using soybean plant images affected by all known plant diseases and insect pests in different growth periods of flowering period, pod bearing period, grain swelling period and mature period, extracting and establishing characteristic parameters of soybean leaves affected by known plant diseases and insect pests in a specific growth period by combining the method in the step 3, converting color leaf images affected by known plant diseases and insect pests in the specific growth period into gray level maps, performing histogram statistics on the gray level maps by using the method in the step 3, finding out a gray level T1 corresponding to the lowest frequency point between the peak values of two gray level parts of a green part of the leaf and a color-changing part affected by the plant diseases and insect pests, and increasing the gray levelSetting the brightness values of all pixels with a threshold T1 as 0, setting the brightness values of more pixels with the gray levels smaller than the threshold T1 as 1, converting the gray level image into a binary image, calculating the actual side length Z1 of each pixel as W1 '/W1 cm according to the pixel number value W1 corresponding to the width of the leaf image and the measurement result W1' cm of the plant corresponding to the leaf image, and setting the actual area of each pixel as Z1 2 Square centimeter, according to the actual side length Z1 centimeter of the pixel and the actual area Z1 of the pixel 2 Square centimeter and the number of pixels corresponding to the extraction result of different parameters of each spot, calculating the mean value of the characteristic parameters of the plant leaf of the known pest type in the specific growth period, including the ratio A1 of the area of the leaf spot of the known pest plant, the number A2 of the leaf spot of the known pest plant, the value A3 of the leaf spot area of the known pest plant, and the value A4 of the circumference of the leaf spot of the known pest plant, in order to further extract the overall morphological characteristics of the leaf affected by the known type of pest, performing histogram statistics on the gray map of the plant affected by the known pest, finding the gray value T1 ' corresponding to the lowest gray level frequency point between the two peak values of the gray level parts of the soybean leaf and the background color, setting the brightness values of all the pixels with the gray level greater than the threshold value T1 ' to be 0, setting the brightness values of all the pixels with the gray level less than T1 ' to be 1, and realizing the binarization processing of the whole soybean leaf, extracting the number of pixels corresponding to the long axis and the short axis of the leaf according to the binary leaf map, and calculating the ratio A5 of the length and the width of the leaf spot of the known plant diseases and insect pests in the specific growth period and the value A6 of the ellipse spot of the known plant diseases and insect pests;
extracting and establishing a library of soybean root characteristic parameters under the influence of known plant diseases and insect pests in a certain specific growth period by combining the method in the step 3, converting a color root image under the influence of the known plant diseases and insect pests in the specific growth period into a gray map, performing histogram statistics on the gray map by using the method in the step 3, finding a gray value T2 corresponding to a lowest gray level frequency point between two gray part peaks of a soybean root and a background color, setting all pixel brightness values with the gray level greater than a threshold T2 as 0, setting all pixel brightness values with the gray level less than T2 as 1, converting the root gray map into a binary map, and converting the root gray map into the binary map according to pixel numerical values W2 and W2 corresponding to the root image widthMeasuring result W2 'cm corresponding to the root system image, calculating actual side length Z2 of each pixel W2'/W2 cm, and actual area of each pixel Z2 2 Square centimeter, according to the actual side length Z2 cm of the pixel and the actual area Z2 of the pixel 2 The method comprises the steps of calculating the mean value of all characteristic parameters of the plant root system affected by known plant diseases and insect pests, including the area value B1 of the plant root system of the known plant diseases and insect pests and the root nodule number value B2 of the plant root system of the known plant diseases and insect pests, calculating the contrast texture characteristic value B3 of the plant root system of the known plant diseases and insect pests according to a formula 1 in the third step, calculating the consistency texture characteristic value B4 of the plant root system of the known plant diseases and insect pests according to a formula 2, calculating the entropy texture characteristic value B5 of the plant root system of the known plant diseases and insect pests according to a formula 3, and calculating the energy texture characteristic value B6 of the plant root system of the known plant diseases and insect pests according to a formula 4;
for the seeds of soybean plants affected by known plant diseases and insect pests in a certain specific growth period, converting a color seed image affected by the known plant diseases and insect pests in the specific growth period into a gray map by combining the method in the step 3, performing histogram statistics on the gray map to find a gray value T3 corresponding to the lowest frequency point between the peak values of two gray parts of a soybean seed part and a background part of a test bench, setting the brightness value of all pixels with the gray value larger than a threshold value T3 as 1, setting the brightness value of most pixels with the gray value smaller than the threshold value T3 as 0, converting the gray map of the seeds into a binary map, calculating the actual side length Z3 ═ W3 '/W3 cm of each pixel according to the pixel number W3 corresponding to the width of the seed image affected by the plant diseases and insect pests and the measurement result W3' cm of the plants corresponding to the root system image, and setting the actual area of each pixel as Z3 2 Square centimeter, according to the actual side length Z3 centimeter of the pixel and the actual area Z3 of the pixel 2 Square centimeter and pixel numbers corresponding to different parameter extraction results of plant seeds affected by diseases and insect pests, and calculating the mean value of characteristic parameters of the known plant seeds affected by the diseases and insect pests in the specific growth period, wherein the mean value comprises the numerical value C1 of the known plant seeds affected by the diseases and insect pests, the area value C2 of the known plant seeds affected by the diseases and insect pests, the girth value C3 of the known plant seeds, the length-width ratio C4 of the known plant seeds affected by the diseases and insect pests, and the known disease and insect pestsDetermining the weight K of characteristic parameters of leaves of each known soybean pest type according to the difference of the influence of different types of pests on the organs of the soybean plants, wherein the grain curvature value C5 of the pest plant and the ellipse value C6 of the known pest plant, and the leaf characteristic parameters of each known soybean pest type 1 Determining the weight K of the root characteristic parameter of each known soybean pest type 2 Determining the weight K of each known soybean pest type seed characteristic parameter 3 While satisfying the normalized characteristic of the weight parameter, i.e. K 1 +K 2 +K 3 1, establishing a pest and disease distinguishing model according to the extracted different organ image characteristic parameter extraction results of the soybean pest and disease types to be identified and the different organ image characteristic parameter extraction results of the known pest and disease types in the database, wherein the distinguishing model is shown as a formula 5:
3. the method for effectively identifying soybean pests and diseases according to claim 2, wherein the method comprises the following steps:
and 5: identifying and identifying the types of the affected soybean plant diseases and insect pests by using plant characteristic parameters affected by different types of known soybean plant diseases and insect pests in different growth periods in the database in the step 4 and known plant disease and pest type distinguishing models in the database, and extracting results a1, a2, a3, a4, a5 and a6 of the characteristic parameters of leaves of a certain plant to be identified; extracting results b1, b2, b3, b4, b5 and b6 of the characteristic parameters of the roots of plants suffering from the plant diseases and insect pests to be identified; and (3) extracting characteristic parameter extraction results c1, c2, c3, c4, c5 and c6 of the seeds of the plants to be identified, bringing the characteristic parameter extraction result of each soybean to be identified affected by the plant diseases and insect pests and the characteristic parameter extraction result of each known soybean affected by the plant diseases and insect pests in the growing period into the discrimination model in the step (4), calculating the P value of the model calculation results of the types of the plant diseases and insect pests to be identified and all the known plant diseases and insect pests according to the discrimination model, wherein the smaller the P value calculation result is, the closer the type of the plant diseases and insect pests to be identified is to the known plant disease and insect pest type corresponding to the P value calculation result is proved, and therefore the identification and identification of the plant diseases and insect pests of the soybean are realized.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011455125.5A CN112668404B (en) | 2020-12-10 | 2020-12-10 | Effective identification method for soybean diseases and insect pests |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011455125.5A CN112668404B (en) | 2020-12-10 | 2020-12-10 | Effective identification method for soybean diseases and insect pests |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112668404A CN112668404A (en) | 2021-04-16 |
| CN112668404B true CN112668404B (en) | 2022-08-30 |
Family
ID=75402591
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011455125.5A Active CN112668404B (en) | 2020-12-10 | 2020-12-10 | Effective identification method for soybean diseases and insect pests |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112668404B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113310989B (en) * | 2021-04-26 | 2024-05-14 | 洛阳农林科学院 | Assessment method for severity of soybean disease green |
| CN116310846B (en) * | 2023-05-19 | 2023-08-29 | 成都大学 | Disease and pest identification method and system based on image identification and data mining |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103559511A (en) * | 2013-11-20 | 2014-02-05 | 天津农学院 | Automatic identification method of foliar disease image of greenhouse vegetable |
| CN207300299U (en) * | 2017-06-30 | 2018-05-01 | 中国农业科学院农业信息研究所 | A kind of soybean diseases and insect pests monitoring car and monitoring system |
| CN108664979A (en) * | 2018-05-10 | 2018-10-16 | 河南农业大学 | The construction method of Maize Leaf pest and disease damage detection model based on image recognition and application |
| CN111401245A (en) * | 2020-03-16 | 2020-07-10 | 吉林农业科技学院 | Crop disease and insect pest detection method, system and equipment based on image |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102334422B (en) * | 2010-07-27 | 2013-02-27 | 中国农业科学院蔬菜花卉研究所 | Machine vision based real-time diagnosis method and system of vegetable leaf diseases |
| US8542939B2 (en) * | 2011-08-04 | 2013-09-24 | Arm Limited | Methods of and apparatus for using tree representations for representing arrays of data elements for encoding and decoding data in data processing systems |
| CN103514459A (en) * | 2013-10-11 | 2014-01-15 | 中国科学院合肥物质科学研究院 | Method and system for identifying crop diseases and pests based on Android mobile phone platform |
| CN105760871A (en) * | 2014-12-14 | 2016-07-13 | 仲恺农业工程学院 | Novel method for identifying plant leaf spot disease resistance |
| TWI656334B (en) * | 2018-02-14 | 2019-04-11 | 國立中山大學 | A system for early detection of orchid pest by hyperspectral imaging techniques |
| CN109063816B (en) * | 2018-08-28 | 2024-02-13 | 武汉古奥基因科技有限公司 | Bean pod number statistics device and method |
| CN109470639A (en) * | 2018-11-07 | 2019-03-15 | 浙江大学 | Diaphania larva based on visible light and near-infrared high light spectrum image-forming and its method for quickly identifying to mulberry leaf damage |
| CN110033015A (en) * | 2019-02-20 | 2019-07-19 | 天津科技大学 | A kind of plant disease detection method based on residual error network |
| CN110188635B (en) * | 2019-05-16 | 2021-04-30 | 南开大学 | Plant disease and insect pest identification method based on attention mechanism and multi-level convolution characteristics |
| CN112052904A (en) * | 2020-09-09 | 2020-12-08 | 陕西理工大学 | Method for identifying plant diseases and insect pests based on transfer learning and convolutional neural network |
-
2020
- 2020-12-10 CN CN202011455125.5A patent/CN112668404B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103559511A (en) * | 2013-11-20 | 2014-02-05 | 天津农学院 | Automatic identification method of foliar disease image of greenhouse vegetable |
| CN207300299U (en) * | 2017-06-30 | 2018-05-01 | 中国农业科学院农业信息研究所 | A kind of soybean diseases and insect pests monitoring car and monitoring system |
| CN108664979A (en) * | 2018-05-10 | 2018-10-16 | 河南农业大学 | The construction method of Maize Leaf pest and disease damage detection model based on image recognition and application |
| CN111401245A (en) * | 2020-03-16 | 2020-07-10 | 吉林农业科技学院 | Crop disease and insect pest detection method, system and equipment based on image |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112668404A (en) | 2021-04-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10568316B2 (en) | Apparatus and methods for in-field data collection and sampling | |
| CN106372592B (en) | A kind of winter wheat planting area calculation method based on winter wheat area index | |
| CN100416590C (en) | Method for automatically identifying field weeds in crop seeding-stage using site and grain characteristic | |
| CN112668404B (en) | Effective identification method for soybean diseases and insect pests | |
| CN114821362B (en) | Multi-source data-based rice planting area extraction method | |
| CN102495005A (en) | Method for diagnosing crop water deficit through hyperspectral image technology | |
| CN109522929A (en) | It is a kind of based on multi-source image fusion natural environment under wheat weeds recognition methods | |
| CN111462223A (en) | Method for identifying soybean and corn planting area in Jianghuai region based on Sentinel-2 image | |
| CN116108318B (en) | Rape nitrogen fertilizer recommended dressing amount calculation method based on unmanned aerial vehicle multispectral image | |
| CN112215090A (en) | Remote sensing rice mapping method integrating phenological knowledge and application thereof | |
| CN116308866B (en) | Rice ear biomass estimation method and system based on canopy reflection spectrum | |
| CN112668406B (en) | Effective screening method for soybean varieties | |
| CN112287886A (en) | Wheat plant nitrogen content estimation method based on hyperspectral image fusion map features | |
| CN114140695B (en) | Prediction method and system for tea tree nitrogen diagnosis and quality index determination based on unmanned aerial vehicle multispectral remote sensing | |
| CN108280410A (en) | One kind being based on binary-coded crops recognition methods and system | |
| CN112345467B (en) | Model for estimating physiological parameters of rice by using remote sensing technology and application thereof | |
| CN116482041B (en) | Rice heading period nondestructive rapid identification method and system based on reflection spectrum | |
| CN114494689A (en) | Identification method of tomato drought stress | |
| CN116258844A (en) | Rapid and accurate identification method for phenotype character of cotton leaf | |
| CN115035423A (en) | Hybrid rice male and female parent identification and extraction method based on unmanned aerial vehicle remote sensing image | |
| CN114047139A (en) | Standard curve color card for representing different nitrogen concentrations of leaves and preparation method and application thereof | |
| Yihang et al. | Automatic recognition of rape seeding emergence stage based on computer vision technology | |
| CN117253145B (en) | Method and system for determining growth situation of broom corn millet and mung bean in different intercropping modes | |
| CN114548220B (en) | Paddy rice glume blooming state detection method based on handheld spectroradiometer | |
| Scott et al. | Estimation of Pyrethrum flower number using digital imagery |
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 |