CN113610965A - Point cloud-based three-dimensional spatial distribution visualization method for chlorophyll content of plants - Google Patents

Point cloud-based three-dimensional spatial distribution visualization method for chlorophyll content of plants Download PDF

Info

Publication number
CN113610965A
CN113610965A CN202110921475.4A CN202110921475A CN113610965A CN 113610965 A CN113610965 A CN 113610965A CN 202110921475 A CN202110921475 A CN 202110921475A CN 113610965 A CN113610965 A CN 113610965A
Authority
CN
China
Prior art keywords
color
chlorophyll content
plant
value
channel
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202110921475.4A
Other languages
Chinese (zh)
Inventor
张慧春
杨琨琪
张萌
边黎明
周宏平
郑加强
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing Forestry University
Original Assignee
Nanjing Forestry University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Nanjing Forestry University filed Critical Nanjing Forestry University
Priority to CN202110921475.4A priority Critical patent/CN113610965A/en
Publication of CN113610965A publication Critical patent/CN113610965A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T17/00Three dimensional [3D] modelling, e.g. data description of 3D objects
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T15/003D [Three Dimensional] image rendering

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Computer Graphics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Geometry (AREA)
  • Software Systems (AREA)
  • Image Processing (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a point cloud-based visualization method for three-dimensional spatial distribution of plant chlorophyll content, which comprises the following steps: collecting color image data of plant leaves; detecting the chlorophyll content of the collected plant leaves by a chlorophyll content measuring instrument; extracting color factors from color image data of plant leaves; performing correlation analysis on the chlorophyll content detected by the chlorophyll content measuring instrument and the extracted color factor, and establishing an optimal regression model of the chlorophyll content; and applying the optimal regression model of the chlorophyll content to the reconstructed plant three-dimensional model to obtain corresponding chlorophyll content values of all points, and realizing three-dimensional spatial distribution visualization of the chlorophyll content of the plant after pseudo-color processing. The invention solves the problem that the chlorophyll content can not be measured rapidly, accurately and nondestructively in large batch in the prior plant phenotype information extraction, and realizes the visualization of the three-dimensional spatial distribution of the chlorophyll content of the plant so as to observe the distribution condition of the chlorophyll content of the plant visually.

Description

Point cloud-based three-dimensional spatial distribution visualization method for chlorophyll content of plants
Technical Field
The invention relates to the field of image analysis, in particular to a point cloud-based visualization method for three-dimensional spatial distribution of chlorophyll content in plants.
Background
When plants are analyzed for physiological and biochemical parameters, it is a common method to collect plant images by computer vision technology and analyze color information in the image data and parameters detected by a physiological and biochemical content measuring instrument to establish a mathematical estimation model. The method for collecting the two-dimensional image by using the visible light camera has the advantages of low cost and easiness in acquisition, and is a common method in the field of computer vision. The two-dimensional plant image can only be used for single-side imaging analysis of plants, but the plants have complex spatial morphological structures, obvious morphological structures and physiological and biochemical changes in the growth process, and the topological structures are generally complex. It is difficult to obtain an accurate measurement value only from two-dimensional image analysis, and complete morphological structure information of the plant, such as leaf area of a curved leaf surface of the plant, organ parameters blocked by a part of branches and leaves, and the like, cannot be obtained. The color information of each organ of the plant is accurately analyzed by establishing the three-dimensional spatial model of the plant, the method can be used for searching the physiological and biochemical parameter change rule of each organ of the plant in the whole growth cycle process of the plant, and has important values for plant fertilization management, phenotype monitoring and pest and disease identification research.
Chlorophyll plays an important role in absorption and utilization of light energy when plants are subjected to photosynthesis. The distribution of chlorophyll content in plant can be used as the basis for plant lack of nutrition or environmental influence. The chlorophyll content of the plant has correlation with the nitrogen content, and the chlorophyll content can be used as an important index for the precise management of the nitrogen fertilizer of the plant. The plant is guided to be fertilized by chlorophyll content, so that not only can the fertilizer waste be effectively reduced, but also the excessive fertilization can be avoided. However, the traditional chlorophyll content measurement can be performed on plants only in a specific time or growth stage, and some plants are divided into leaves and plants in vitro by breaking, picking, cutting and other ways to perform destructive measurement, so that the workload is large, the efficiency is low, and the measurement can be performed only on a single point on a single leaf.
Disclosure of Invention
The invention aims to solve the technical problem of providing a point cloud-based visualization method for three-dimensional spatial distribution of plant chlorophyll content, which aims at overcoming the defects of the prior art, solves the problem that the chlorophyll content cannot be measured rapidly, accurately and nondestructively in large batch in the existing plant phenotype information extraction, and realizes visualization of the three-dimensional spatial distribution of the plant chlorophyll content so as to observe the distribution condition of the plant chlorophyll content visually.
In order to achieve the technical purpose, the technical scheme adopted by the invention is as follows:
a visualization method of three-dimensional spatial distribution of plant chlorophyll content based on point cloud comprises the following steps:
s1, calibrating color information parameters of the visible light camera and collecting color image data of the plant leaves;
s2, detecting the chlorophyll content of the collected plant leaves by a chlorophyll content measuring instrument;
s3, processing the collected color image data of the plant leaves, and extracting color factors;
s4, performing correlation analysis on the chlorophyll content detected by the chlorophyll content measuring instrument and the color factors extracted after the processing, and establishing an optimal regression model of the chlorophyll content;
s5, collecting color image data of the plant under multiple visual angles;
s6, extracting feature point information in the plant color image collected under multiple visual angles by using a scale and rotation invariance algorithm;
s7, carrying out feature point matching on the feature points in each plant color image in the S6 through proximity search;
s8, obtaining a plant three-dimensional model with color information through a motion recovery structure algorithm;
and S9, applying the optimal regression model of chlorophyll content established in the S4 to the plant three-dimensional model reconstructed in the S8 to obtain corresponding chlorophyll content values of all points, and performing pseudo-color processing to realize three-dimensional space distribution visualization of chlorophyll content of the plant.
As a further improved technical solution of the present invention, the visible light camera color information parameter calibration in step S1 specifically includes the following steps:
s11, collecting images of a red, green and blue three-primary-color card by a visible light camera, wherein RGB color channel values of the red color card are (255, 0, 0) respectively, RGB color channel values of the green color card are (0, 255, 0) respectively, and RGB color channel values of the blue color card are (0, 0, 255) respectively;
s12, extracting an R channel from a red color card image, extracting a G channel from a green color card image, extracting a B channel from a blue color card image, calculating the value of an average pixel point under the R channel, recording the value as Y _ R, calculating the value of the average pixel point under the G channel, recording the value as Y _ G, calculating the value of the average pixel point under the B channel, recording the value as Y _ B, wherein the value calculation formula of the average pixel point is as follows:
Figure BDA0003207569100000021
wherein Y _ x is the value of the average pixel point under the x channel, fx(i, j) is the value of i row and j column under x channel;
s13, calculating the average pixel point value and the theoretical value of each channel obtained in the step S12 to obtain a color correction proportionality coefficient Z _ R under the R channel, a color correction proportionality coefficient Z _ G under the G channel and a color correction proportionality coefficient Z _ B under the B channel, wherein the calculation function of the color correction proportionality coefficient is as follows:
Figure BDA0003207569100000022
wherein, Z _ x is a color correction proportion coefficient under an x channel, and Y _ x is a value of an average pixel point under the x channel;
and S14, applying the color correction scale coefficients of the channels obtained in the step S13 to all the color images collected by the visible light camera to realize color correction.
As a further improved technical solution of the present invention, the detecting of the chlorophyll content of the collected plant leaf by the chlorophyll content measuring instrument in step S2 specifically includes:
s21, carrying out calibration treatment on the chlorophyll content measuring instrument;
and S22, measuring the chlorophyll content of three different positions on the collected single plant leaf, and taking the average value of the chlorophyll content of the three different positions as the chlorophyll content of the plant leaf.
As a further improved technical solution of the present invention, the step S3 specifically includes:
s31, firstly, converting the collected color image of the plant leaf into a gray image;
s32, carrying out binarization processing on the transformed gray level image to realize the segmentation between the plant leaves and the background;
s33, if the plant leaf image after the binarization processing has certain noise, carrying out image noise reduction processing;
s34, carrying out mask processing on the image acquired in the step S1 and the binary image subjected to noise reduction processing in the step S33 through an image mask algorithm, wherein the mask processing refers to: taking the white pixel region in the binary image subjected to noise reduction processing in the step S33 as a region of interest ROI, performing bit operation on the region of interest ROI and the image acquired in the step S1, obtaining a self numerical value after the bit operation is performed, and changing the image numerical values of other regions into 0, so as to completely separate the plant leaf part from the background in the image;
s35, converting the plant leaf image subjected to mask processing in the step S34 into different color spaces, wherein the different color spaces comprise an RGB color space, a La B color space and an HSV color space, and extracting color factors from the different color spaces, wherein the color factors comprise R, G, B, H, S, V, L, a and B; calculating the number of pixel points occupied by the plant leaves and the sum of the color factors under a single channel, and dividing the sum of the color factors under the single channel by the number of the pixel points occupied by the plant leaves to serve as the numerical value of the color factors of the plant leaves under the channel;
Figure BDA0003207569100000031
wherein Fx(i, j) is the value of i row and j column under x channel, S is the calculated number of pixel points occupied by plant leaves, FxThe value of the color factor of the plant leaf under the x channel is shown; the x channel refers to R, G, B color factor in RGB color space, H, S, V color factor in HSV color space, or L, a, b color factor in La b color space, respectively;
calculating a color factor combination value, the color factor combination value comprising
Figure BDA0003207569100000032
G2
S36, normalizing the color factor value and the color factor combination value obtained from each channel in the step S35, and converting the value in the original value interval [0, 255] into the range [0, 1] interval.
As a further improved technical solution of the present invention, the step S4 specifically includes:
and (4) carrying out linear and nonlinear polynomial regression model establishment on the color factors and the color factor combination values extracted from different channels in different color spaces obtained in the step (S3) and the chlorophyll content values of the plant leaves detected by the chlorophyll content measuring instrument in the step (S2), and selecting a model which shows the best fitting performance by utilizing various model evaluation indexes as an optimal regression model of the chlorophyll content.
As a further improved technical solution of the present invention, the step S5 specifically includes:
the visible light camera is fixed, plants are placed on the tray to rotate, or the plants are fixed, the visible light camera is placed on the platform to rotate by taking the plants as the center, image acquisition is carried out on each plant at an interval of a rotation angle of 18 degrees, and color image data of 20 plants are acquired in one circle.
As a further improved technical solution of the present invention, the step S9 specifically includes:
a91, applying the optimal regression model of the chlorophyll content in the step S4 to the normalized color information of the plant three-dimensional model reconstructed in the step S8, so that the normalized color information of the three-dimensional model is converted into an estimated value of the chlorophyll content;
a92, wherein the distribution range of the chlorophyll content estimated value in the step A91 is [0, 100], the distribution range of the chlorophyll content estimated value is extended to [0, 255], and the full gray value interval distribution condition of the chlorophyll content of the plant is obtained;
and A93, performing pseudo-color treatment on the processed plant three-dimensional model in the step A92 to enable the chlorophyll content of the plant to present a visual effect according to spatial distribution.
The invention has the beneficial effects that:
a point cloud-based visualization method for three-dimensional spatial distribution of plant chlorophyll content is characterized in that a regression model among chlorophyll content, plant color factors and color factor combinations is established and a plant three-dimensional model is reconstructed, the regression model is applied to the point cloud and subjected to pseudo-color processing, visualization of the three-dimensional spatial distribution of the plant chlorophyll content is achieved, the problem that the chlorophyll content cannot be measured in a large batch, fast, accurate and lossless mode in the existing plant phenotype information extraction is solved, visualization of the three-dimensional spatial distribution of the plant chlorophyll content is achieved, the distribution condition of the plant chlorophyll content is observed visually, and the method can be used as a guide basis for plant growth monitoring, yield estimation, accurate fertilization management and aging degree judgment; the chlorophyll content of the plant can be measured at any time, and the chlorophyll content of any point of the plant leaf can be measured and visualized.
Drawings
FIG. 1 is a schematic flow chart of a visualization method for three-dimensional spatial distribution of plant chlorophyll content based on point cloud.
Fig. 2 is a schematic view of a plant leaf image processing flow in the point cloud-based visualization method for three-dimensional spatial distribution of plant chlorophyll content.
Fig. 3 is a schematic diagram of a chlorophyll content space visualization processing flow in the point cloud-based plant chlorophyll content three-dimensional space three-dimensional distribution visualization method of the present invention.
Fig. 4 (a) is a schematic diagram of an arabidopsis thaliana three-dimensional point cloud model in an application example of chlorophyll content space visualization in the point cloud-based plant chlorophyll content three-dimensional space three-dimensional distribution visualization method of the present invention.
Fig. 4 (b) is a schematic diagram of the visualization of the spatial distribution of the chlorophyll content in arabidopsis thaliana in the application example of the spatial visualization of the chlorophyll content in the point cloud-based three-dimensional spatial distribution of the chlorophyll content in the plant material.
Fig. 4 (c) is a schematic diagram of a three-dimensional point cloud model of osmanthus fragrans leaves in an application example of spatial visualization of chlorophyll content in a point cloud-based three-dimensional spatial distribution visualization method of plant chlorophyll content.
Fig. 4 (d) is a schematic diagram of visualization of spatial distribution of chlorophyll content in osmanthus tree leaves in an application example of spatial visualization of chlorophyll content in a point cloud-based visualization method of three-dimensional spatial distribution of chlorophyll content in plants.
Fig. 5 (a) is a grayscale diagram of fig. 4 (a).
Fig. 5 (b) is a grayscale diagram of fig. 4 (b).
Fig. 5 (c) is a grayscale diagram of fig. 4 (c).
Fig. 5 (d) is a grayscale diagram of fig. 4 (d).
Detailed Description
The following further description of embodiments of the invention is made with reference to the accompanying drawings:
as shown in fig. 1, a schematic flow chart is provided for an embodiment of a method for visualizing three-dimensional spatial distribution of plant chlorophyll content based on point cloud, and the image analysis method includes:
and S1, calibrating the color information parameters of the visible light camera and acquiring color image data of the plant leaves.
And S2, detecting the chlorophyll content of the collected plant leaves by using a chlorophyll content measuring instrument.
S3, inputting the collected color image data of the plant leaf into an image processing program, processing the plant leaf image to extract image color factors including R, G, B, G G,
Figure BDA0003207569100000051
H. S, V, L, a, B, wherein R, G, B, H, S, V, L, a, B belong to color factors under a single channel, G,
Figure BDA0003207569100000052
The combined color factor may also be referred to as a color factor combination.
And S4, establishing a regression model by using the chlorophyll content of the plant leaves detected by the chlorophyll content measuring instrument and the color factors extracted after the treatment.
S5, carrying out rotary shooting (the visible light camera is still, the plant is placed on the tray to rotate or the plant is still, the visible light camera is placed on the platform to rotate by taking the plant as the center), collecting color image data of the plant under multiple viewing angles, carrying out image collection by taking 18 degrees as a rotation angle interval, and collecting the color image data of 20 plants in one circle.
And S6, extracting feature point information in the plant color image collected under multiple visual angles by using a Scale-invariant feature transform (SIFT) algorithm.
And S7, carrying out feature point matching on the feature points in each plant color image of the S6 through proximity search.
S8, obtaining the plant three-dimensional model with color information through a Structure From Motion (SFM) algorithm.
And S9, applying the regression model established in the S4 to the plant three-dimensional model reconstructed in the S8 to obtain corresponding chlorophyll content values of all points, stretching the chlorophyll content values to a range of [0, 255], and performing pseudo-color treatment to realize three-dimensional spatial distribution visualization of the chlorophyll content of the plant.
The color image data of the plant leaves in the step S1 is acquired by the visible light camera, so that the conditions of the light environment, the position of the visible light camera and the like are consistent, the color parameters of the visible light camera are calibrated before the acquisition, and the calibration of the color parameters of the visible light camera comprises the following steps:
s11 first collects images of three primary color cards of red, green, and blue by a visible light camera, wherein RGB color channel values of the three color cards of red, green, and blue are (255, 0, 0), (0, 255, 0), (0, 0, 255), respectively.
S12 extracts an R channel from the collected red color card image, extracts a G channel from the collected green color card image, extracts a B channel from the collected blue color card image, calculates a value Y _ R of an average pixel point under the R channel, calculates a value Y _ G of an average pixel point under the G channel, and calculates a value Y _ B of an average pixel point under the B channel, where the average pixel point calculation formula is as follows:
Figure BDA0003207569100000061
wherein Y _ x is the value of the average pixel point under the x channel, fx(i, j) is the value of i row and j column under x channel.
S13 calculates the average pixel value and the theoretical value in each channel obtained in S12 to obtain a color correction scaling factor Z _ R in the R channel, a color correction scaling factor Z _ G in the G channel, and a color correction scaling factor Z _ B in the B channel, where the calculation function of the color correction scaling factor is as follows:
Figure BDA0003207569100000062
wherein, Z _ x is the color correction scale factor under the x channel, and Y _ x is the value of the average pixel under the x channel.
S14 applies the respective channel color correction scale coefficients obtained in S13 described above to all subsequent color images for color correction.
When the chlorophyll content measuring instrument in the step S2 is used for collection, the measuring instrument is calibrated, after the calibration is completed, the plant leaf (avoiding veins) is inserted and the measuring probe is closed, three different positions on the plant leaf are searched for, and the operation is repeated, and the average value of three measurement values is taken as the chlorophyll content of the leaf collected in the step S1.
As shown in fig. 2, the image processing algorithm for processing the plant leaf image to extract the image color factor in step S3 includes the steps of:
s31, firstly, converting the color image of the plant leaf into a gray image, wherein the conversion function is as follows:
Figure BDA0003207569100000071
wherein I is a gray value, fx(i, j) is the value of i row and j column under x channel.
And S32, performing binarization processing on the gray level image converted in the S31 to realize the division between the plant leaves and the background.
S33, if the plant leaf image after the threshold processing of S32 has a certain noise, the image noise reduction processing is performed by setting an appropriate convolution kernel and an opening operation algorithm in image morphology, where the opening operation function is as follows:
Figure BDA0003207569100000072
wherein A is an image set, B is a convolution kernel, D is an expansion operation, and E is a corrosion operation.
S34 performs mask processing on the image in S1 and the binary image after S33 noise reduction processing by using an image mask algorithm, where the mask processing is to perform bit operation on the white pixel region in the binary image after S33 processing as a region of interest roi (region of interest) and the image in S1, and when the value in the image in S1 is bit operated on the region of interest, the obtained value is still a self value, and the values of other partial images become 0, so that the plant leaf part in the image is completely separated from the background.
S35 transforms the plant leaf image after mask processing in S34 into different color spaces, including RGB (Red, Green, Blue) color space, La × b (Lab color space) color space, HSV (Hue, Satura formation, Value) color space, as shown in table 1, extracting color factors from different color spaces and their combinations. And calculating the sum of the number of the pixel points occupied by the plant leaves and the color factor under a single channel, and dividing the sum of the color factor under the single channel by the number of the pixel points occupied by the plant leaves to serve as the numerical value of the color factor of the plant leaves under the channel.
Figure BDA0003207569100000073
Wherein Fx(i, j) is the value of i row and j column under x channel, S is the calculated number of pixel points occupied by plant leaves, FxThe x channel refers to R, G, B color factor in RGB color space, H, S, V color factor in HSV color space, and L, a, b in La b color space, respectively, which are the values of the color factor under the x channel.
R, G, B, converting the color components representing green degree into another 4 common color factors (i.e. color factors)
Figure BDA0003207569100000074
G2
Figure BDA0003207569100000075
G2Is a combined color factor, which may also be referred to as a color factor combination).
Table 1:
Figure BDA0003207569100000081
s36, normalizing the color factor value and the color factor combination value obtained from each channel in S35, and converting the value in the original value interval [0, 255] into the range [0, 1 ].
Wherein the result of step S3 is not obtained in step S4Carrying out linear and nonlinear polynomial regression model establishment on color factors and color factor combinations extracted from different channels in the same color space and the chlorophyll content values of the plant leaves detected by the chlorophyll content measuring instrument in the S2; specifically, a single or a plurality of color factors are randomly selected from all the color factors in table 1, and the color factors and the chlorophyll content value of the plant leaf detected by the chlorophyll content measuring instrument are subjected to correlation analysis, so that a linear and nonlinear polynomial regression model is established. And selecting the model with the best fitting performance as the optimal regression model by using various model evaluation indexes. Wherein the multiple model evaluation indexes comprise root mean square error RMSE and decision coefficient R2
Figure BDA0003207569100000082
The formula is a unified expression of the established regression model, wherein Y is the chlorophyll content obtained by the regression model, and X is1、X2…XnIs the color factor and color factor combination of each channel in n different color spaces, i.e. n color factors in Table 1, n is greater than or equal to 1 and less than or equal to 13, w1、w2…wnIs a weight coefficient, A1、A2…AnThe base number of the logarithm is B, the weight coefficient is B, and a, B and c … n are degree terms, and the regression model of the color factors under different color spaces is shown in Table 2. The RMSE stated in table 2 represents the average prediction error compared to the measured values, with lower values and higher accuracy. R2The percentage of the measurement variance, which represents the model fitting effect and is interpreted by the algorithm model, has a value range of [0, 1]],R2The larger the model, the better the model fit.
Figure BDA0003207569100000091
Figure BDA0003207569100000092
Wherein y isrealThe real value clipped by the hand-held chlorophyll measuring instrument is free of dimensional units; y ispredThe predicted value of the multi-color factor correlation model is a dimensionless unit; m is the number of data, and the unit is a group; y ismeanThe predicted value of the multi-color factor correlation model is the average value of the predicted values of all the multi-color factor correlation models, and is dimensionless.
Table 2:
Figure BDA0003207569100000093
Figure BDA0003207569100000101
the numbers 1-5 in Table 2 are used to establish a linear regression model, and the numbers 6-15 are used to establish a non-linear regression model, from which it can be seen that the regression model 14 constructed by the color factors lg (G), R, G, B, G/R, G/B and the regression model 15 constructed by the color factors lg (G), R, G, B, G/(R + B) determine the coefficient R2Both up to 0.73, but the model 14 incorporating the G/R and G/B color factors had less error (RMSE 2.16 ═ R<2.21), so regression model 14: y ═ 8.51 × lg (G) +11.68 × R-26.48 × G +18.30 × B +2.81 × G/R +3.85 × G/B +40 fitted best to the SPAD values, with the most significant regression, as the model fitted for optimal chlorophyll content. The G/R and G/B linear regression models prove that the green degree of the plant leaves can be better reflected by the color factor of G/(R + B), and the effects of the quadratic term regression models with the numbers of 7-10 and the logarithmic term regression models with the numbers of 11-15 prove that the G/R and the G/B can be separately adjusted as two parameters, so that the relative green degree of the plant leaves can be better fitted.
The step S5 of collecting color image data of a plant under multiple viewing angles is implemented by placing a visible light camera on a platform, rotating the plant on a tray, or rotating the plant around the plant. The precise control device rotates in a circle, image acquisition is carried out on each plant at an interval of 18-degree rotation angle, and color image data of 20 plants are acquired in one circle.
In the step S6, the plant image feature detection uses the SIFT algorithm, the SIFT algorithm calculates the position information (x, y) of the feature point through gaussian filters of different sizes, and simultaneously provides a Descriptor information, and in a grid histogram around the feature point, each histogram includes a gradient direction, so as to obtain a multi-dimensional feature vector, thereby realizing the feature detection of the plant image.
In the step S7, the distance of the nearest neighbor is set to be d1 by the proximity search algorithm, then the distance between the second nearest matching pairs is found to be d2, if the ratio of the two distances d1 and d2 is less than a threshold, an acceptable matching pair can be determined, a sampling consistency algorithm (RANSC) is used to calculate a base matrix for the matching points, and matching pairs which do not satisfy the base matrix are eliminated. The feature point matching principle function is as follows:
Figure BDA0003207569100000102
wherein f isd1Is a characteristic point, fd2Is another feature point on the image, fnnIs the nearest neighbor feature vector, and F (J) is image J.
As shown in fig. 3, the visualization of the three-dimensional spatial distribution of chlorophyll content in S9 includes the steps of:
s91 applies the regression model of chlorophyll content in S4 to the normalized color information of the three-dimensional plant model reconstructed in S8, so that the normalized color information of the three-dimensional plant model is converted into an estimated value of chlorophyll content.
S92, setting the range of the chlorophyll content estimated value in the S91 as 0, 100, extending the distribution range of the chlorophyll content estimated value to 0, 255, and obtaining the full gray value interval distribution condition of the chlorophyll content of the plant.
And S93, performing pseudo-color processing on the processed plant three-dimensional model in the S92 to enable the chlorophyll content of the plant to present a visual effect according to spatial distribution.
r(i,j=TR[f(i,j)]
g(i,j=TG[f(i,j)]
b(i,j=TB[f(i,j)]
Where r (i, j, g (i, j, b) (i, j are the values of the three components of the pseudo-color image red, green, and blue, respectively), f (i, j is the gray level of the original image, TR、TB、TGRespectively, representing a linear mapping of gray scale levels to the R, G, B primaries. Wherein T isRThe linear mapping relationship means that the gray level is lower than HmaxThe/2's are each mapped to the darkest red color; gray value at Hmax/2~3HmaxBetween/4, the brightness of red increases linearly with gray level; gray scale value of 3Hmax/4~HmaxIn between, the red color remains unchanged at the brightest level; t isB、TCAnd the above-mentioned TRThe mapping relation is the same, wherein HmaxIs the maximum gray value 255.
Fig. 4 is a schematic diagram of a chlorophyll content space visualization application example (arabidopsis thaliana and osmanthus tree leaves) in the point cloud-based plant chlorophyll content three-dimensional space three-dimensional distribution visualization method. FIG. 4 (a) is a schematic diagram of the Arabidopsis thaliana three-dimensional point cloud model constructed according to the steps S5-S8. Fig. 4 (b) is a schematic diagram illustrating the spatial distribution of the chlorophyll content in arabidopsis thaliana established in step S9. Fig. 4 (c) is a schematic diagram of the three-dimensional point cloud model of the osmanthus tree leaves constructed according to the steps S5-S8.
Fig. 4 (d) is a schematic diagram of visualization of the chlorophyll content spatial distribution of the osmanthus tree leaves established according to step S9. Fig. 5 is a grayscale diagram of fig. 4.
According to the invention, plant leaf images are processed, color factors and combinations of the plant leaves are extracted, a regression model between the color factors and the combinations of the plant leaves and chlorophyll content obtained by a chlorophyll content measuring instrument is established, the regression model is applied to a plant three-dimensional model, chlorophyll content of all point clouds is obtained, and spatial bright distribution of the chlorophyll content of the plant is obtained after normalization and pseudo-color processing. The chlorophyll content of the whole plant is displayed on the three-dimensional structure of the plant in different colors, and the change can be seen before the change is seen by naked eyes, and quantitative analysis is carried out.
The point cloud-based visualization method for three-dimensional spatial distribution of the chlorophyll content of the plants provides guidance basis for growth monitoring, yield estimation, accurate fertilization management and aging degree judgment of the plants, and has very important significance.
The scope of the present invention includes, but is not limited to, the above embodiments, and the present invention is defined by the appended claims, and any alterations, modifications, and improvements that may occur to those skilled in the art are all within the scope of the present invention.

Claims (7)

1. A visualization method of three-dimensional spatial distribution of plant chlorophyll content based on point cloud is characterized in that: the method comprises the following steps:
s1, calibrating color information parameters of the visible light camera and collecting color image data of the plant leaves;
s2, detecting the chlorophyll content of the collected plant leaves by a chlorophyll content measuring instrument;
s3, processing the collected color image data of the plant leaves, and extracting color factors;
s4, performing correlation analysis on the chlorophyll content detected by the chlorophyll content measuring instrument and the color factors extracted after the processing, and establishing an optimal regression model of the chlorophyll content;
s5, collecting color image data of the plant under multiple visual angles;
s6, extracting feature point information in the plant color image collected under multiple visual angles by using a scale and rotation invariance algorithm;
s7, carrying out feature point matching on the feature points in each plant color image in the S6 through proximity search;
s8, obtaining a plant three-dimensional model with color information through a motion recovery structure algorithm;
and S9, applying the optimal regression model of chlorophyll content established in the S4 to the plant three-dimensional model reconstructed in the S8 to obtain corresponding chlorophyll content values of all points, and performing pseudo-color processing to realize three-dimensional space distribution visualization of chlorophyll content of the plant.
2. The method for visualizing the three-dimensional spatial distribution of the chlorophyll content in plants based on point cloud as claimed in claim 1, wherein: the visible light camera color information parameter calibration in step S1 specifically includes the following steps:
s11, collecting images of a red, green and blue three-primary-color card by a visible light camera, wherein RGB color channel values of the red color card are (255, 0, 0) respectively, RGB color channel values of the green color card are (0, 255, 0) respectively, and RGB color channel values of the blue color card are (0, 0, 255) respectively;
s12, extracting an R channel from a red color card image, extracting a G channel from a green color card image, extracting a B channel from a blue color card image, calculating the value of an average pixel point under the R channel, recording the value as Y _ R, calculating the value of the average pixel point under the G channel, recording the value as Y _ G, calculating the value of the average pixel point under the B channel, recording the value as Y _ B, wherein the value calculation formula of the average pixel point is as follows:
Figure FDA0003207569090000011
wherein Y _ x is the value of the average pixel point under the x channel, fx(i, j) is the value of i row and j column under x channel;
s13, calculating the average pixel point value and the theoretical value of each channel obtained in the step S12 to obtain a color correction proportionality coefficient Z _ R under the R channel, a color correction proportionality coefficient Z _ G under the G channel and a color correction proportionality coefficient Z _ B under the B channel, wherein the calculation function of the color correction proportionality coefficient is as follows:
Figure FDA0003207569090000012
wherein, Z _ x is a color correction proportion coefficient under an x channel, and Y _ x is a value of an average pixel point under the x channel;
and S14, applying the color correction scale coefficients of the channels obtained in the step S13 to all the color images collected by the visible light camera to realize color correction.
3. The method for visualizing the three-dimensional spatial distribution of the chlorophyll content in plants based on point cloud as claimed in claim 1, wherein: the step S2 of detecting the chlorophyll content of the collected plant leaves by the chlorophyll content measuring instrument specifically includes:
s21, carrying out calibration treatment on the chlorophyll content measuring instrument;
and S22, measuring the chlorophyll content of three different positions on the collected single plant leaf, and taking the average value of the chlorophyll content of the three different positions as the chlorophyll content of the plant leaf.
4. The method for visualizing the three-dimensional spatial distribution of the chlorophyll content in plants based on point cloud as claimed in claim 1, wherein: the step S3 specifically includes:
s31, firstly, converting the collected color image of the plant leaf into a gray image;
s32, carrying out binarization processing on the transformed gray level image to realize the segmentation between the plant leaves and the background;
s33, if the plant leaf image after the binarization processing has certain noise, carrying out image noise reduction processing;
s34, carrying out mask processing on the image acquired in the step S1 and the binary image subjected to noise reduction processing in the step S33 through an image mask algorithm, wherein the mask processing refers to: taking the white pixel region in the binary image subjected to noise reduction processing in the step S33 as a region of interest ROI, performing bit operation on the region of interest ROI and the image acquired in the step S1, obtaining a self numerical value after the bit operation is performed, and changing the image numerical values of other regions into 0, so as to completely separate the plant leaf part from the background in the image;
s35, converting the plant leaf image subjected to mask processing in the step S34 into different color spaces, wherein the different color spaces comprise an RGB color space, a La B color space and an HSV color space, and extracting color factors from the different color spaces, wherein the color factors comprise R, G, B, H, S, V, L, a and B; calculating the number of pixel points occupied by the plant leaves and the sum of the color factors under a single channel, and dividing the sum of the color factors under the single channel by the number of the pixel points occupied by the plant leaves to serve as the numerical value of the color factors of the plant leaves under the channel;
Figure FDA0003207569090000021
wherein Fx(i, j) is the value of i row and j column under x channel, S is the calculated number of pixel points occupied by plant leaves, FxThe value of the color factor of the plant leaf under the x channel is shown; the x channel refers to R, G, B color factor in RGB color space, H, S, V color factor in HSV color space, or L, a, b color factor in La b color space, respectively;
calculating a color factor combination value, the color factor combination value comprising
Figure FDA0003207569090000022
G2
S36, normalizing the color factor value and the color factor combination value obtained from each channel in the step S35, and converting the value in the original value interval [0, 255] into the range [0, 1] interval.
5. The method for visualizing the three-dimensional spatial distribution of the chlorophyll content in plants based on point cloud as claimed in claim 1, wherein: the step S4 specifically includes:
and (4) carrying out linear and nonlinear polynomial regression model establishment on the color factors and the color factor combination values extracted from different channels in different color spaces obtained in the step (S3) and the chlorophyll content values of the plant leaves detected by the chlorophyll content measuring instrument in the step (S2), and selecting a model which shows the best fitting performance by utilizing various model evaluation indexes as an optimal regression model of the chlorophyll content.
6. The method for visualizing the three-dimensional spatial distribution of the chlorophyll content in plants based on point cloud as claimed in claim 1, wherein: the step S5 specifically includes:
the visible light camera is fixed, plants are placed on the tray to rotate, or the plants are fixed, the visible light camera is placed on the platform to rotate by taking the plants as the center, image acquisition is carried out on each plant at an interval of a rotation angle of 18 degrees, and color image data of 20 plants are acquired in one circle.
7. The method for visualizing the three-dimensional spatial distribution of the chlorophyll content in plants based on point cloud as claimed in claim 1, wherein: the step S9 specifically includes:
a91, applying the optimal regression model of the chlorophyll content in the step S4 to the normalized color information of the plant three-dimensional model reconstructed in the step S8, so that the normalized color information of the three-dimensional model is converted into an estimated value of the chlorophyll content;
a92, wherein the distribution range of the chlorophyll content estimated value in the step A91 is [0, 100], the distribution range of the chlorophyll content estimated value is extended to [0, 255], and the full gray value interval distribution condition of the chlorophyll content of the plant is obtained;
and A93, performing pseudo-color treatment on the processed plant three-dimensional model in the step A92 to enable the chlorophyll content of the plant to present a visual effect according to spatial distribution.
CN202110921475.4A 2021-08-11 2021-08-11 Point cloud-based three-dimensional spatial distribution visualization method for chlorophyll content of plants Pending CN113610965A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110921475.4A CN113610965A (en) 2021-08-11 2021-08-11 Point cloud-based three-dimensional spatial distribution visualization method for chlorophyll content of plants

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110921475.4A CN113610965A (en) 2021-08-11 2021-08-11 Point cloud-based three-dimensional spatial distribution visualization method for chlorophyll content of plants

Publications (1)

Publication Number Publication Date
CN113610965A true CN113610965A (en) 2021-11-05

Family

ID=78308282

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110921475.4A Pending CN113610965A (en) 2021-08-11 2021-08-11 Point cloud-based three-dimensional spatial distribution visualization method for chlorophyll content of plants

Country Status (1)

Country Link
CN (1) CN113610965A (en)

Similar Documents

Publication Publication Date Title
Kolhar et al. Plant trait estimation and classification studies in plant phenotyping using machine vision–A review
Narvekar et al. Grape leaf diseases detection & analysis using SGDM matrix method
CN112766155A (en) Deep learning-based mariculture area extraction method
CN112287886A (en) Wheat plant nitrogen content estimation method based on hyperspectral image fusion map features
Lang et al. Detection of chlorophyll content in Maize Canopy from UAV Imagery
CN110610438B (en) Crop canopy petiole included angle calculation method and system
CN109870419B (en) Method for predicting nitrogen, phosphorus and potassium content of black soil by adopting aviation hyperspectral data
Lin et al. Estimating litchi flower number using a multicolumn convolutional neural network based on a density map
Zhang et al. Multi-phenotypic parameters extraction and biomass estimation for lettuce based on point clouds
Liu et al. UAV multispectral images for accurate estimation of the maize LAI considering the effect of soil background
CN113570538A (en) Blade RGB image skewed distribution parameter information acquisition and analysis method
CN117058522A (en) Remote sensing earth surface ecological change detection method integrating spectral neural network and leaf area index
Ibaraki et al. Image analysis for plants: basic procedures and techniques
Yao et al. Image-based plant nutrient status analysis: An overview
CN113610965A (en) Point cloud-based three-dimensional spatial distribution visualization method for chlorophyll content of plants
Akila et al. Automation in plant growth monitoring using high-precision image classification and virtual height measurement techniques
CN116416523A (en) Machine learning-based rice growth stage identification system and method
Di et al. The research on the feature extraction of sunflower leaf rust characteristics based on color and texture feature
Wang et al. Study on estimation method of plant leaf area based on image processing technology
Hall et al. A method for extracting detailed information from high resolution multispectral images of vineyards
Rony et al. BottleNet18: Deep Learning-Based Bottle Gourd Leaf Disease Classification
CN112949607A (en) Wetland vegetation feature optimization and fusion method based on JM Relief F
Chang et al. Improved deep learning-based approach for real-time plant species recognition on the farm
Mangla et al. Statistical growth prediction analysis of rice crop with pixel-based mapping technique
Zhou et al. Maize-pas: Automated maize phenotyping analysis software using deep learning

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