Crop close planting resistant variety breeding method and device based on unmanned aerial vehicle time sequence image

Abstract

The invention provides a crop close-planting-resistant variety breeding method and device based on an unmanned aerial vehicle time sequence image, and relates to the technical field of agricultural informationization, and the method comprises the steps: obtaining a time sequence image collected by an unmanned aerial vehicle for a crop planting region, and planting a plurality of varieties of crops in different densities in the crop planting region; based on the time sequence image, extracting a canopy development stage corresponding to the crop of each variety and an intermediate feature corresponding to the canopy development stage, and quantifying a canopy development rate difference index of the crop of each variety under different densities by using the intermediate feature; and according to the density tolerance index corresponding to the crop of each variety and the correlation between the density tolerance index and the canopy development rate difference index under different densities, screening out a target intermediate feature so as to realize the breeding of the crop density-tolerant variety by using the target intermediate feature. According to the method, canopy development rate difference indexes of crop varieties under different planting densities are effectively quantified, and the breeding efficiency and reliability of the crop close-planting-resistant varieties are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of agricultural information technology, and in particular to a method and device for breeding crop varieties resistant to dense planting based on unmanned aerial vehicle time-series images. Background Art

[0002] The rate of crop canopy development largely determines its efficiency in absorbing light, thus affecting yield. When breeding varieties that tolerate dense planting, it is crucial to understand the growth dynamics of crops under high-density planting conditions. Therefore, it is important to quickly and accurately quantify the development rate of crops during critical periods, define their early vitality, mid-term vitality, and maturity phenotypes, and analyze the mechanism by which these phenotypic characteristics affect their tolerance to dense planting, which is of great significance for optimizing breeding strategies and breeding ideal varieties that tolerate dense planting.

[0003] Traditional methods for breeding varieties that tolerate dense planting mainly rely on manual field statistics of crop development parameters at different densities, such as plant height, yield, and number of pods. This method is inefficient, has large subjective errors, and is difficult to effectively quantify growth rates. Especially when there are many breeding materials, real-time statistics of growth information become more difficult, posing a severe challenge to high-throughput variety breeding. UAV phenotyping technology uses drones equipped with a variety of sensors to monitor plants. Previous studies have mainly focused on collecting data once during each growth period and combining machine learning to build an inversion model. However, it is unable to reflect the growth and development dynamics of crop varieties under different planting densities, affecting the reliability of breeding varieties that tolerate dense planting. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a method and device for breeding crop varieties resistant to dense planting based on UAV time-series images, which effectively quantifies the canopy development rate difference index of crop varieties under different planting densities, and significantly improves the efficiency and reliability of breeding crop varieties resistant to dense planting.

[0005] In a first aspect, an embodiment of the present invention provides a method for breeding crop varieties resistant to dense planting based on time-series images of unmanned aerial vehicles, comprising:

[0006] Obtain time-series images collected by drones for crop planting areas. Multiple varieties of crops are planted at different densities in the crop planting areas. The time-series images include RGB images, multispectral orthophoto images, and elevation images corresponding to each variety of crops.

[0007] Based on the time series images, the canopy development stage and the intermediate features corresponding to the canopy development stage of each crop variety are extracted, so as to quantify the canopy development rate difference index of each crop variety at different densities using the intermediate features.

[0008] According to the correlation between the density tolerance index corresponding to each variety of crops and its canopy development rate difference index under different densities, the target intermediate characteristics are screened out, so as to utilize the target intermediate characteristics to realize the breeding of crop varieties resistant to dense planting.

[0009] In one embodiment, based on the time series images, the canopy development stage corresponding to each variety of crops and the intermediate features corresponding to the canopy development stage are extracted, including:

[0010] Segment the time series images to obtain RGB sub-images, multispectral orthophoto sub-images, and elevation sub-images of each crop variety;

[0011] For each crop variety, the canopy coverage, plant height and leaf area index of the crop variety are extracted based on the RGB sub-image, multispectral orthophoto sub-image and elevation sub-image of the crop variety to fit the canopy development time series curve of the crop variety.

[0012] According to the canopy development time series curve of the crop variety, the canopy development stage corresponding to the crop variety and the intermediate features corresponding to the canopy development stage are extracted.

[0013] In one embodiment, based on the RGB sub-image, the multispectral orthophoto sub-image, and the elevation sub-image of the crop of the variety, the canopy coverage, plant height, and leaf area index of the crop of the variety are extracted to fit the canopy development time series curve of the crop of the variety, including:

[0014] Extract the canopy coverage of the crop of this variety based on the RGB sub-image;

[0015] and, extracting the plant height of the crop of the variety based on the elevation sub-image;

[0016] and, extracting texture features of the crop of the variety based on the RGB sub-image, extracting a vegetation index of the crop of the variety based on the RGB sub-image and the multispectral orthophoto sub-image, and outputting a leaf area index of the crop of the variety based on the vegetation index and the texture features through a pre-trained leaf area index estimation model;

[0017] For the canopy coverage, plant height and leaf area index of this variety of crops, P splines were fitted respectively to obtain the canopy coverage time series curve, plant height time series curve and leaf area index time series curve as the canopy development time series curve of this variety of crops.

[0018] In one embodiment, the training step of the leaf area index estimation model comprises:

[0019] The measured values ​​of the leaf area index in the designated sub-area within the crop planting area are used as training labels, and the texture characteristics and vegetation index of the crop variety planted in the designated sub-area are used as model inputs to train the Bi-LSTM model based on bidirectional hidden states to obtain a leaf area index estimation model.

[0020] The bidirectional hidden state includes a forward hidden state and a backward hidden state. The forward hidden state represents the leaf area index from the past to the present, and the backward hidden state represents the leaf area index from the future to the present.

[0021] In one embodiment, based on the canopy development time series curve of the crop of the variety, extracting the canopy development stage corresponding to the crop of the variety and the intermediate features corresponding to the canopy development stage include:

[0022] Based on the plant height changes represented by the plant height time series curve of the crop variety, the canopy development stage corresponding to the crop variety is divided into the early stage, the middle stage and the mature stage;

[0023] For each stage, the intermediate features characterizing the canopy development rate at that stage were extracted from the canopy coverage time series curve, plant height time series curve, and leaf area index time series curve.

[0024] In one embodiment, the intermediate features are used to quantify the canopy development rate difference index of each crop variety at different densities, including:

[0025] For each intermediate feature, determining the ratio of the intermediate features of the crop of the variety at different densities to obtain an intermediate feature ratio;

[0026] Each intermediate characteristic ratio is used as the canopy development rate difference index of the crop variety.

[0027] In one embodiment, according to the correlation between the density tolerance index corresponding to each variety of crops and the canopy development rate difference index under different densities, the target intermediate features are screened out, including:

[0028] For each variety of crops, the Pearson correlation coefficient between the density tolerance index corresponding to the variety of crops and the ratio of the intermediate characteristics of the variety of crops is determined, and the target intermediate characteristics are screened out from the intermediate characteristics in descending order of the Pearson correlation coefficient.

[0029] In a second aspect, an embodiment of the present invention further provides a device for selecting and breeding crop varieties resistant to dense planting based on time-series images of unmanned aerial vehicles, comprising:

[0030] An image acquisition module is used to acquire time-series images collected by drones for crop planting areas. Multiple varieties of crops are planted at different densities in the crop planting area. The time-series images include RGB images, multispectral orthophoto images, and elevation images corresponding to each variety of crops.

[0031] A development rate difference quantification module is used to extract the canopy development stage corresponding to each variety of crops and the intermediate features corresponding to the canopy development stage based on time series images, so as to quantify the canopy development rate difference index of each variety of crops at different densities using the intermediate features;

[0032] The variety breeding module is used to screen out the target intermediate characteristics based on the correlation between the density tolerance index corresponding to each variety of crops and its canopy development rate difference index under different densities, so as to use the target intermediate characteristics to realize the breeding of crop varieties that are resistant to dense planting.

[0033] In a third aspect, an embodiment of the present invention further provides an electronic device, comprising a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement any one of the methods provided in the first aspect.

[0034] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement any one of the methods provided in the first aspect.

[0035] The embodiment of the present invention provides a method and device for breeding crop varieties resistant to dense planting based on unmanned aerial vehicle time-series images. First, the time-series images collected by the unmanned aerial vehicle for the crop planting area are obtained. Multiple varieties of crops are planted at different densities in the crop planting area. The time-series images include RGB images, multispectral orthophoto images, and elevation images corresponding to each variety of crops. Then, based on the time-series images, the canopy development stage corresponding to each variety of crops and the intermediate features corresponding to the canopy development stage are extracted to quantify the canopy development rate difference index of each variety of crops at different densities using the intermediate features. Finally, according to the correlation between the density tolerance index corresponding to each variety of crops and the canopy development rate difference index at different densities, the target intermediate features are screened out to realize the breeding of crop varieties resistant to dense planting using the target intermediate features. The above method extracts the canopy development stage of crops and the corresponding intermediate features through the time-series data collected by the unmanned aerial vehicle, quantifies the canopy development rate difference index of crops at different planting densities, and combines the density tolerance index corresponding to the crops with the correlation between the canopy development rate difference index at different densities to realize the breeding of crop varieties resistant to dense planting, thereby significantly improving the efficiency and reliability of the breeding of crop varieties resistant to dense planting.

[0036] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0037] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0039] Figure 1 A schematic diagram of a process for selecting crop varieties resistant to dense planting based on time-series images of unmanned aerial vehicles provided in an embodiment of the present invention;

[0040] Figure 2 An overall flow chart of a method for selecting crop varieties resistant to dense planting based on time-series images of unmanned aerial vehicles provided in an embodiment of the present invention;

[0041] Figure 3 A flow chart of a dynamic model-based density analysis provided by an embodiment of the present invention;

[0042] Figure 4 A schematic diagram of a time series curve provided by an embodiment of the present invention;

[0043] Figure 5 A correlation between a canopy rate difference and a density tolerance index provided by an embodiment of the present invention;

[0044] Figure 6 A schematic diagram of breeding dense planting tolerant crop varieties based on intermediate characteristics provided by an embodiment of the present invention;

[0045] Figure 7 A schematic diagram of the structure of a crop variety breeding device for dense planting resistance based on time-series images of unmanned aerial vehicles provided in an embodiment of the present invention;

[0046] Figure 8 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described in combination with the embodiments below. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] At present, the traditional method of breeding varieties resistant to dense planting mainly relies on manually counting the development parameters of crops in the field. This method is inefficient and has large subjective errors. There are few studies on the collection of continuous crop images and the construction of dynamic models in combination with time series algorithms. It is difficult to effectively quantify the growth rate and the differences of crops under different densities, which brings severe challenges to high-throughput variety breeding. Based on this, the present invention provides a method and device for breeding varieties resistant to dense planting of crops based on drone time series images, which effectively quantifies the canopy development rate difference index of crop varieties under different planting densities, and significantly improves the efficiency and reliability of breeding varieties resistant to dense planting of crops.

[0049] To facilitate understanding of this embodiment, firstly, a method for selecting crop varieties resistant to dense planting based on drone time-series images disclosed in an embodiment of the present invention is described in detail, see Figure 1 The flowchart of a method for selecting crop varieties resistant to dense planting based on time-series images of unmanned aerial vehicles is shown, and the method mainly includes the following steps S102 to S106:

[0050] Step S102, obtaining time-series images collected by the drone for the crop planting area.

[0051] In the crop planting area, multiple varieties of crops are planted at different densities, and the time series images include RGB images, multispectral orthophoto images, and elevation images corresponding to each variety of crops. In one example, a drone is equipped with a digital camera and a multispectral camera to take images of the crop planting area in a vertical photography manner. Since the emergence of crops, a flight is performed every N (such as 5) days, and the captured images are spliced ​​to generate time series RGB images, multispectral orthophoto images, and elevation images.

[0052] Step S104, based on the time series images, extracting the canopy development stage corresponding to each variety of crops and the intermediate features corresponding to the canopy development stage, so as to quantify the canopy development rate difference index of each variety of crops at different densities using the intermediate features.

[0053] Among them, the canopy development stage is divided into the early stage, the middle stage and the mature stage. The early stage refers to the stage from the emergence of the crop to the first plant height, the middle stage refers to the stage from the first plant height to the second plant height, and the mature stage refers to the stage after the crop reaches the second plant height. The first plant height is related to the second plant height. For example, if the second plant height is the maximum plant height, the first plant height is 25% of the maximum plant height. The intermediate characteristics of the early stage include: the time point when the maximum plant height is 25%, the canopy coverage and leaf area index at this time point, and the average change rate of the early plant height, canopy coverage and leaf area index; the intermediate characteristics of the middle stage include: the maximum values ​​of plant height, canopy coverage and leaf area index, and the average change rate of the three in the middle stage; the intermediate characteristics of the mature stage include: the average change rate of plant height, canopy coverage and leaf area index in the mature stage. The canopy development rate difference index is the ratio of the intermediate characteristics of the early, middle and mature stages of each variety at different densities.

[0054] In one example, the canopy coverage, plant height and leaf area index of each crop variety were first extracted based on the time series images to fit the canopy development time series curve of each crop variety. The canopy development stage and the intermediate characteristics corresponding to the canopy development stage were then further extracted using the canopy development time series curve. Finally, the ratio of the intermediate characteristics of the early, middle and mature stages of each variety at different densities was calculated to quantify the canopy development rate difference index of each crop variety at different densities.

[0055] Step S106, based on the correlation between the density tolerance index corresponding to each variety of crops and the canopy development rate difference index under different densities, the target intermediate characteristics are screened out to achieve the breeding of crop varieties resistant to dense planting using the target intermediate characteristics.

[0056] In one example, the correlation between the density tolerance index and the canopy development rate difference index can be characterized by the Pearson correlation coefficient. The higher the value of the Pearson correlation coefficient, the higher the possibility that the corresponding intermediate characteristic will be used as the target intermediate characteristic required for the breeding of crop varieties tolerant to dense planting.

[0057] The method for selecting and breeding crop varieties resistant to dense planting based on unmanned aerial vehicle time-series images provided in an embodiment of the present invention extracts the canopy development stage of the crop and its corresponding intermediate characteristics through the time-series data collected by the unmanned aerial vehicle, quantifies the canopy development rate difference index of the crop under different planting densities, and combines the corresponding density tolerance index of the crop with the correlation between its canopy development rate difference index under different densities to realize the selection and breeding of crop varieties resistant to dense planting, thereby significantly improving the efficiency and reliability of the selection and breeding of crop varieties resistant to dense planting.

[0058] In specific implementation, the embodiment of the present invention utilizes drone phenotyping technology, combined with time series prediction and dynamic models, to analyze the influence mechanism of canopy development rate on density tolerance, and realize high-throughput breeding of crop varieties tolerant to dense planting. Through the high-throughput advantage of drones, the growth and development dynamics of crop varieties under different planting densities are recorded, the canopy development rate is accurately quantified, and combined with statistical analysis methods, correlation analysis of crop variety tolerance is performed, providing a technical basis for high-throughput breeding of breeding materials tolerant to dense planting.

[0059] See also Figure 2 The overall flow chart of a method for selecting crop varieties resistant to dense planting based on UAV time-series images is shown in FIG. Figure 2 The method includes: time series data collection, including drone digital image collection, drone multispectral image collection, and leaf area index measurement of some varieties; image and processing, including image stitching, segmentation, and removal of ground background; feature extraction, including plant height, canopy coverage, vegetation index, texture feature extraction, and leaf area index extraction based on vegetation index and texture feature using Bi-LSTM (Bidirectional Long Short Term Memory); dynamic model construction and intermediate feature extraction, including early vitality features (i.e., intermediate features of the early stage), mid-term vitality features (i.e., intermediate features of the mid-term stage), and maturity features (i.e., intermediate features of the mature stage); correlation relationship of density tolerance index, including correlation coefficient and SHAP (SHapley Additive exPlanations) algorithm. For further information, see Figure 3 The figure shows a flow chart of density tolerance analysis based on a dynamic model. The intermediate features are extracted based on the fitted canopy coverage time series curve, plant height time series curve, and leaf area index time series curve. The density tolerance index and canopy rate differences are combined, and the Pearson correlation coefficient analysis is used to select important intermediate features to achieve density tolerance analysis.

[0060] exist Figure 2 and Figure 3 On the basis of this, an embodiment of the present invention proposes a method for selecting crop varieties resistant to dense planting based on drone time series images. The crop phenotypic parameters are extracted through time series images and algorithms with high throughput, and the crop development rate under different densities is quantified, in order to improve the efficiency and reliability of the selection of crop varieties resistant to dense planting. Specifically, an embodiment of the present invention provides a specific implementation method of a method for selecting crop varieties resistant to dense planting based on drone time series images.

[0061] For the aforementioned step S102, the embodiment of the present invention provides an implementation method for obtaining time-series images collected by a drone for a crop planting area. Specifically:

[0062] The acquired data includes time series images and ground data at designated sub-areas within the crop planting area, which is also the measured value of the leaf area index. In one example, a drone equipped with a digital camera and a multispectral camera takes images of the crop planting area in a vertical photography manner. Since the emergence of the crop seedlings, a flight is carried out every 5 days, and the images taken are spliced ​​to generate time series RGB images, multispectral orthophotos, and elevation images. The ground data is measured by LAI-2200C to measure the measured value of the leaf area index of the designated sub-area.

[0063] For the aforementioned step S104, the embodiment of the present invention further provides an implementation method for extracting the canopy development stage corresponding to each variety of crops and the intermediate features corresponding to the canopy development stage based on the time series image, see the following steps 1A to 1C:

[0064] Step 1A, segmenting the time series image to obtain RGB sub-images, multispectral orthophoto sub-images, and elevation sub-images of each crop variety.

[0065] In one example, QGIS was used to generate a vector file containing geographic location information for each variety, and the spliced ​​time-series images were segmented to obtain RGB sub-images, multispectral orthophoto sub-images, and elevation sub-images of each variety of crops.

[0066] Step 1B, for each variety of crops, based on the RGB sub-image, multispectral orthophoto sub-image, and elevation sub-image of the crop, extract the canopy coverage, plant height, and leaf area index of the crop of the variety to fit the canopy development time series curve of the crop of the variety. In the specific implementation, it includes:

[0067] (1) Extracting the canopy coverage of the crop of the variety based on the RGB sub-image. In one example, based on the segmented RGB sub-image, the soil background is removed using the threshold method, and the canopy coverage of each variety on the date of the drone flight is calculated by dividing the vegetation pixels by the total pixels.

[0068] (2) Extracting the plant height of the crop of the variety based on the elevation sub-image. In one example, based on the segmented elevation sub-image, the percentile method was used to extract the plant height, with the ninety-ninth percentile as the upper boundary and the first percentile as the lower boundary, to obtain the plant height of each variety on all drone flight dates.

[0069] (3) Extract the texture features of the crop of the variety based on the RGB sub-image, extract the vegetation index of the crop of the variety based on the RGB sub-image and the multispectral orthophoto sub-image, and output the leaf area index of the crop of the variety based on the vegetation index and texture features through the pre-trained leaf area index estimation model. Among them, the texture features include mean (MEA), variance (VAR), homogeneity (HOM), contrast (CON), difference (DIS), entropy (ENT), second-order moment (SEC) and correlation (COR). In one example, the gray level co-occurrence matrix method is used in Python to extract texture features from the RGB sub-image; and the DN (Digital Number) average of each band in the RGB sub-image and the multispectral orthophoto sub-image is extracted to calculate the vegetation index; the feature data such as the time series vegetation index of each variety is input into the leaf area index estimation model, and the model will output the corresponding leaf area index prediction value. Finally, the leaf area index of each variety on all UAV flight dates is obtained.

[0070] Before executing (3), it is necessary to pre-build a leaf area index estimation model. In specific implementation, the measured leaf area index value at the designated sub-region within the crop planting area can be used as a training label, and the texture features and vegetation index of the crop of the variety planted in the designated sub-region can be used as model input to train the Bi-LSTM model based on the bidirectional hidden state to obtain a leaf area index estimation model; wherein the bidirectional hidden state includes a forward hidden state and a backward hidden state, the forward hidden state represents the leaf area index from the past to the present, and the backward hidden state represents the leaf area index from the future to the present. In one example, a part of the soybean plots with measured leaf area index values ​​is used as a training set, and an estimation model is established using the improved Bi-LSTM algorithm. The model input includes time series vegetation index, texture features, etc. extracted from RGB sub-images and multispectral orthophoto sub-images. A bidirectional hidden state is added to the existing unidirectional LSTM model, wherein the forward hidden state represents the leaf area index information from the past to the present, and the backward hidden state represents the leaf area index information from the future to the present. Bi-LSTM selectively memorizes and forgets information through the mechanism of memory cells, input gates, forget gates and output gates in a bidirectional processing mode, thereby capturing the dynamic relationship between features such as vegetation index and leaf area index.

[0071] (4) P-splines are fitted for the canopy coverage, plant height, and leaf area index of the crop variety, respectively, to obtain the canopy coverage time series curve, plant height time series curve, and leaf area index time series curve as the canopy development time series curve of the crop variety. This step can also be called dynamic model construction. Among them, P-spline is a semi-parametric model used to fit the trait measurements at regular time intervals. Based on the plant height, canopy coverage, and leaf area index data of all flight drone dates, Python is used to fit the p-spline to obtain the time series curve in days during the growth period of each variety, which is used to characterize the canopy development dynamics of the crop. Such as Figure 4 A schematic diagram of a time series curve is shown, and the time series curve includes a canopy coverage time series curve, a plant height time series curve, and a leaf area index time series curve.

[0072] Step 1C, based on the canopy development time series curve of the crop variety, extract the canopy development stage corresponding to the crop variety and the intermediate features corresponding to the canopy development stage.

[0073] In the specific implementation, the canopy development stage corresponding to the crop variety can be divided into early stage, middle stage and mature stage based on the plant height changes represented by the plant height time series curve of the crop variety; then for each stage, the intermediate characteristics representing the canopy development rate in that stage are extracted from the canopy coverage time series curve, the plant height time series curve and the leaf area index time series curve.

[0074] First, the growth vitality of crops at different stages is extracted. Including: plant height is an intuitive indicator to measure crop growth and development. According to the change of plant height, crop canopy development can be divided into three stages: early, middle and mature. Early refers to the period from seedling emergence to 25% of the maximum plant height; middle refers to the period from this point to the maximum plant height; mature refers to the period after the maximum plant height is reached.

[0075] Then, the intermediate features are extracted to characterize the development status of crops at different stages. Including: plant height, canopy coverage, leaf area index are key indicators reflecting the lateral and vertical growth rate and photosynthesis efficiency of crops within a certain period of time. By analyzing the time series curves of plant height, canopy coverage, and leaf area index of each variety, the intermediate features characterizing the crop development rate at each stage can be extracted. The characteristics of the early stage include: the time point when the maximum plant height reaches 25%, the canopy coverage and leaf area index at this time point, and the average change rate of early plant height, canopy coverage and leaf area index. The characteristics of the mid-stage include: the maximum values ​​of plant height, canopy coverage and leaf area index, and the average change rate of these three in the mid-term. The characteristics of the mature stage include: the average change rate of plant height, canopy coverage and leaf area index in the mature period. For details, see the definitions of intermediate features in different stages shown in Table 1:

[0076] Table 1 Definition of intermediate features at different stages

[0077]

[0078]

[0079] For the aforementioned step S104, an embodiment of the present invention also provides an implementation method for using intermediate features to quantify the canopy development rate difference index of each variety of crops at different densities. For each intermediate feature, the ratio between the intermediate features of the variety of crops at different densities can be determined to obtain the intermediate feature ratio; then each intermediate feature ratio is used as the canopy development rate difference index of the variety of crops.

[0080] In one example, when the crop density was increased, conditions such as radiation interception and nutrient competition changed significantly, which in turn affected the growth rate. In order to quantify the difference in canopy development rate of each variety at high and low densities, the ratio of the intermediate characteristics of each variety at different densities in the early, middle and mature stages was calculated. The calculation formula is as follows:

[0081] Traits Variation (TV) = Trait high density / Trait low density ;

[0082] Among them, Traits Variation (TV) represents the canopy development rate, Traits high density Represents the intermediate characteristics under high planting density, Trait low density Represents intermediate characteristics at low planting density.

[0083] For the aforementioned step S106, an embodiment of the present invention provides a specific implementation method for screening out target intermediate features based on the correlation between the density tolerance index corresponding to each variety of crops and its canopy development rate difference index under different densities, including: for each variety of crops, determining the Pearson correlation coefficient between the density tolerance index corresponding to the variety of crops and the ratio of the intermediate features of the variety of crops, and screening out the target intermediate features from the intermediate features in descending order of the Pearson correlation coefficient.

[0084] Among them, the yield per unit area after increasing the planting area is a direct indicator to measure the density tolerance performance of crop varieties. Therefore, the ratio of the yield per unit area of ​​each variety under high and low planting density is used as the density tolerance index of the variety. The calculation formula of the density tolerance index is as follows:

[0085] Density tolerance (DT) = Yield high density / Yield low density ;

[0086] Among them, Density tolerance (DT) represents the density tolerance index, Yield high density Represents the yield per unit area under high planting density. low density Represents the yield per unit area at low planting density.

[0087] First, the correlation analysis between canopy development rate and density tolerance index is carried out. Including: there are often differences in the development rates of crop varieties under high and low densities, which ultimately leads to different yields per unit area under high and low densities. The DT and TV calculated in the above steps define the difference in growth rate and yield per unit area. To better understand the relationship between them, the Pearson correlation coefficient is calculated to measure the linear relationship between the intermediate characteristic ratio and the density tolerance index. At the same time, the importance of these intermediate characteristic differences is evaluated, and how the changes in the intermediate characteristics after increasing the planting density will lead to increased yields will be explored. This will provide guidance for parsing the relationship between the development rate and the density tolerance index at different stages, such as Figure 5 Shown is a correlation between canopy rate difference and density tolerance index.

[0088] Then, through the correlation coefficient, the target intermediate features that are highly correlated with the density resistance coefficient (DT) at each stage can be obtained. Figure 6 A schematic diagram of breeding crop varieties tolerant to dense planting based on intermediate characteristics is shown in FIG. Figure 6 LAI is given in tPH25 The effect of combining LAI-mid and CC-mature in breeding densely planted varieties. tPH25 The performance of the combined breeding of dense planting tolerant varieties was the strongest. With the increase of planting density, almost all soybean varieties with rapid increase in LAI in the early stage and maintaining a high LAI growth rate in the middle stage eventually achieved higher yield per unit area, confirming that the increase in yield depends on the increase in LAI in the early and middle stages. There is no universal linear relationship between the intermediate trait CC-mature and DT at the maturity stage. However, when compared with LAI tPH25 When combined with LAI-mid, a similar pattern emerged: those varieties that increased LAI in early and mid-season, while also having higher CC-mature at maturity, significantly increased yield per unit area. In contrast, if a variety increased LAI in early season, but had lower CC-mature at maturity, its yield increase would be very limited.

[0089] In summary, the method for selecting crop varieties resistant to dense planting based on UAV time-series images provided in the embodiment of the present invention will quantify the differences in crop development rates under different densities based on time series images, so as to achieve efficient selection of varieties resistant to dense planting. Specifically, as an emerging intelligent breeding method, drone phenotyping can quickly cover large areas of farmland and capture subtle crop growth states. The embodiment of the present invention utilizes the high-throughput advantage of drone phenotyping to realize the construction of a dynamic model based on image-derived features, effectively quantifies the differences in growth rates of different varieties under high and low densities, and combines statistical means to analyze the relationship between growth rate and density tolerance, providing new ideas and methods for the selection of crop varieties resistant to dense planting.

[0090] Based on the above embodiments, the present invention provides a crop variety breeding device based on drone time-series images, see Figure 7 The schematic diagram of the structure of a crop variety breeding device based on UAV time-series images is shown, and the device mainly includes the following parts:

[0091] The image acquisition module 702 is used to acquire time-series images collected by the drone for the crop planting area. Multiple varieties of crops are planted at different densities in the crop planting area. The time-series images include RGB images, multispectral orthophoto images, and elevation images corresponding to each variety of crops.

[0092] The development rate difference quantification module 704 is used to extract the canopy development stage corresponding to each variety of crops and the intermediate features corresponding to the canopy development stage based on the time series images, so as to quantify the canopy development rate difference index of each variety of crops at different densities using the intermediate features;

[0093] The variety breeding module 706 is used to screen out target intermediate characteristics according to the correlation between the density tolerance index corresponding to each variety of crops and its canopy development rate difference index under different densities, so as to use the target intermediate characteristics to realize the breeding of crop varieties that are resistant to dense planting.

[0094] The device for selecting and breeding crop varieties resistant to dense planting based on unmanned aerial vehicle time-series images provided in an embodiment of the present invention extracts the canopy development stage of crops and their corresponding intermediate characteristics through time-series data collected by unmanned aerial vehicles, quantifies the canopy development rate difference index of crops under different planting densities, and combines the corresponding density tolerance index of the crop with the correlation between its canopy development rate difference index under different densities to realize the selection and breeding of crop varieties resistant to dense planting, thereby significantly improving the efficiency and reliability of the selection and breeding of crop varieties resistant to dense planting.

[0095] In one embodiment, the growth rate difference quantification module 704 is specifically used to:

[0096] Segment the time series images to obtain RGB sub-images, multispectral orthophoto sub-images, and elevation sub-images of each crop variety;

[0097] For each crop variety, the canopy coverage, plant height and leaf area index of the crop variety are extracted based on the RGB sub-image, multispectral orthophoto sub-image and elevation sub-image of the crop variety to fit the canopy development time series curve of the crop variety.

[0098] According to the canopy development time series curve of the crop variety, the canopy development stage corresponding to the crop variety and the intermediate features corresponding to the canopy development stage are extracted.

[0099] In one embodiment, the growth rate difference quantification module 704 is specifically used to:

[0100] Extract the canopy coverage of the crop of this variety based on the RGB sub-image;

[0101] and, extracting the plant height of the crop of the variety based on the elevation sub-image;

[0102] and, extracting texture features of the crop of the variety based on the RGB sub-image, extracting a vegetation index of the crop of the variety based on the RGB sub-image and the multispectral orthophoto sub-image, and outputting a leaf area index of the crop of the variety based on the vegetation index and the texture features through a pre-trained leaf area index estimation model;

[0103] For the canopy coverage, plant height and leaf area index of this variety of crops, P splines were fitted respectively to obtain the canopy coverage time series curve, plant height time series curve and leaf area index time series curve as the canopy development time series curve of this variety of crops.

[0104] In one embodiment, a model training module is further included, which is used to:

[0105] The measured values ​​of the leaf area index in the designated sub-area within the crop planting area are used as training labels, and the texture characteristics and vegetation index of the crop variety planted in the designated sub-area are used as model inputs to train the Bi-LSTM model based on bidirectional hidden states to obtain a leaf area index estimation model.

[0106] The bidirectional hidden state includes a forward hidden state and a backward hidden state. The forward hidden state represents the leaf area index from the past to the present, and the backward hidden state represents the leaf area index from the future to the present.

[0107] In one embodiment, the growth rate difference quantification module 704 is specifically used to:

[0108] Based on the plant height changes represented by the plant height time series curve of the crop variety, the canopy development stage corresponding to the crop variety is divided into the early stage, the middle stage and the mature stage;

[0109] For each stage, the intermediate features characterizing the canopy development rate at that stage were extracted from the canopy coverage time series curve, plant height time series curve, and leaf area index time series curve.

[0110] In one embodiment, the growth rate difference quantification module 704 is specifically used to:

[0111] For each intermediate feature, determining the ratio of the intermediate features of the crop of the variety at different densities to obtain an intermediate feature ratio;

[0112] Each intermediate characteristic ratio is used as the canopy development rate difference index of the crop variety.

[0113] In one embodiment, the variety breeding module 706 is specifically used to:

[0114] For each variety of crops, the Pearson correlation coefficient between the density tolerance index corresponding to the variety of crops and the ratio of the intermediate characteristics of the variety of crops is determined, and the target intermediate characteristics are screened out from the intermediate characteristics in descending order of the Pearson correlation coefficient.

[0115] The device provided in the embodiment of the present invention has the same implementation principle and technical effects as those of the aforementioned method embodiment. For the sake of brief description, for matters not mentioned in the device embodiment, reference may be made to the corresponding contents in the aforementioned method embodiment.

[0116] An embodiment of the present invention provides an electronic device. Specifically, the electronic device includes a processor and a storage device. The storage device stores a computer program, and when the computer program is executed by the processor, it executes the method described in any one of the above-mentioned implementation methods.

[0117] Figure 8 A structural diagram of an electronic device provided in an embodiment of the present invention, the electronic device 100 includes: a processor 80, a memory 81, a bus 82 and a communication interface 83, wherein the processor 80, the communication interface 83 and the memory 81 are connected via the bus 82; the processor 80 is used to execute an executable module stored in the memory 81, such as a computer program.

[0118] The memory 81 may include a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk memory. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 83 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used.

[0119] The bus 82 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 8 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0120] Among them, the memory 81 is used to store programs, and the processor 80 executes the program after receiving the execution instruction. The method executed by the device for flow process definition disclosed in any embodiment of the above-mentioned embodiments of the present invention can be applied to the processor 80 or implemented by the processor 80.

[0121] The processor 80 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 80. The above processor 80 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in the embodiments of the present invention can be directly embodied as a hardware decoding processor to be executed, or the hardware and software modules in the decoding processor can be executed. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 81, and the processor 80 reads the information in the memory 81 and completes the steps of the above method in combination with its hardware.

[0122] The computer program product of the readable storage medium provided in the embodiment of the present invention includes a computer-readable storage medium storing program code, and the instructions included in the program code can be used to execute the methods described in the previous method embodiments. The specific implementation can be referred to the previous method embodiments, which will not be repeated here.

[0123] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0124] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for breeding crop varieties resistant to dense planting based on UAV time-series images, characterized in that: include: Acquire time-series images collected by a drone for a crop planting area, wherein multiple varieties of crops are planted at different densities in the crop planting area, and the time-series images include RGB images, multispectral orthophoto images, and elevation images corresponding to each variety of crops; Based on the time series images, extracting the canopy development stage corresponding to each of the varieties of crops and the intermediate features corresponding to the canopy development stage, so as to quantify the canopy development rate difference index of each of the varieties of crops at different densities using the intermediate features; According to the correlation between the density tolerance index corresponding to each variety of crops and the canopy development rate difference index under different densities, the target intermediate characteristics are screened out to utilize the target intermediate characteristics to realize the breeding of crop varieties that are resistant to dense planting.

2. The method for selecting crop varieties resistant to dense planting based on UAV time-series images according to claim 1, characterized in that: Based on the time series images, extracting the canopy development stage corresponding to each variety of crops and the intermediate features corresponding to the canopy development stage, including: Segmenting the time series images to obtain RGB sub-images, multispectral orthophoto sub-images, and elevation sub-images of each crop variety; For each of the varieties of crops, based on the RGB sub-image, the multispectral orthophoto sub-image, and the elevation sub-image of the variety of crops, the canopy coverage, plant height, and leaf area index of the variety of crops are extracted to fit a canopy development time series curve of the variety of crops; According to the canopy development time series curve of the crop variety, the canopy development stage corresponding to the crop variety and the intermediate features corresponding to the canopy development stage are extracted.

3. The method for selecting crop varieties resistant to dense planting based on UAV time-series images according to claim 2, characterized in that: Based on the RGB sub-image, the multispectral orthophoto sub-image, and the elevation sub-image of the crop of the variety, the canopy coverage, plant height, and leaf area index of the crop of the variety are extracted to fit the canopy development time series curve of the crop of the variety, including: Extracting the canopy coverage of the crop of the variety based on the RGB sub-image; and, extracting the plant height of the crop of the variety based on the elevation sub-image; and, extracting texture features of the crop of the variety based on the RGB sub-image, extracting a vegetation index of the crop of the variety based on the RGB sub-image and the multispectral orthophoto sub-image, and outputting a leaf area index of the crop of the variety based on the vegetation index and the texture features through a pre-trained leaf area index estimation model; For the canopy coverage, plant height and leaf area index of this variety of crops, P splines are fitted respectively to obtain a canopy coverage time series curve, a plant height time series curve and a leaf area index time series curve as the canopy development time series curve of this variety of crops.

4. The method for selecting crop varieties resistant to dense planting based on UAV time-series images according to claim 3 is characterized in that: The training step of the leaf area index estimation model comprises: Taking the measured value of the leaf area index at the designated sub-region within the crop planting area as a training label, and taking the texture features and the vegetation index of the crop of the variety planted in the designated sub-region as model inputs, training a Bi-LSTM model based on a bidirectional hidden state, and obtaining a leaf area index estimation model; The bidirectional hidden state includes a forward hidden state and a backward hidden state, the forward hidden state represents the leaf area index from the past to the present, and the backward hidden state represents the leaf area index from the future to the present.

5. The method for selecting crop varieties resistant to dense planting based on UAV time-series images according to claim 3, characterized in that: According to the canopy development time series curve of the crop of the variety, the canopy development stage corresponding to the crop of the variety and the intermediate features corresponding to the canopy development stage are extracted, including: Based on the plant height changes represented by the plant height time series curve of the crop of the variety, the canopy development stage corresponding to the crop of the variety is divided into an early stage, a middle stage and a mature stage; For each stage, intermediate features characterizing the canopy development rate at that stage are extracted from the canopy coverage time series curve, the plant height time series curve, and the leaf area index time series curve.

6. The method for selecting crop varieties resistant to dense planting based on UAV time-series images according to claim 1, characterized in that: The intermediate features are used to quantify the canopy development rate difference index of each crop variety at different densities, including: For each of the intermediate characteristics, determining the ratio of the intermediate characteristics of the crop of the variety at different densities to obtain an intermediate characteristic ratio; Each of the intermediate characteristic ratios is used as a canopy development rate difference index for the crop variety.

7. The method for selecting crop varieties resistant to dense planting based on UAV time-series images according to claim 6, characterized in that: According to the correlation between the density tolerance index corresponding to each variety of crops and the canopy development rate difference index under different densities, the target intermediate features are screened out, including: For each variety of crops, the Pearson correlation coefficient between the density tolerance index corresponding to the variety of crops and the ratio of the intermediate characteristics of the variety of crops is determined, and the target intermediate characteristics are screened out from the intermediate characteristics in descending order of the Pearson correlation coefficients.

8. A device for selecting and breeding crop varieties resistant to dense planting based on time-series images from unmanned aerial vehicles, characterized in that: include: An image acquisition module is used to acquire time-series images collected by a drone for a crop planting area, wherein multiple varieties of crops are planted at different densities in the crop planting area, and the time-series images include RGB images, multispectral orthophoto images, and elevation images corresponding to each variety of crops; A development rate difference quantification module is used to extract the canopy development stage corresponding to each of the crop varieties and the intermediate features corresponding to the canopy development stage based on the time series images, so as to quantify the canopy development rate difference index of each of the crop varieties at different densities using the intermediate features; The variety breeding module is used to screen out target intermediate characteristics based on the correlation between the density tolerance index corresponding to each variety of crops and the canopy development rate difference index under different densities, so as to use the target intermediate characteristics to realize the breeding of crop varieties that are resistant to dense planting.

9. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer executable instructions that can be executed by the processor, and the processor executes the computer executable instructions to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the method according to any one of claims 1 to 7.

Images