Remote sensing monitoring method and electronic equipment for crop growth period in plastic greenhouse

Abstract

The present invention provides a remote sensing monitoring method and electronic equipment for crop growth period in a plastic greenhouse, the method comprising: obtaining remote sensing images of the plastic greenhouse where the crops are located; extracting object features based on the remote sensing images; the object features include spectral features, shape features and texture features; inputting the spectral features, shape features and texture features into a preset crop growth period classification model; wherein the crop growth period classification model is obtained by training the spectral features, shape features and texture features of different landforms; judging the crop growth period based on the prediction results of the crop growth period classification model. The present invention is used to solve the defects of the prior art that the traditional manual survey method is used for monitoring the growth period information of crops in the greenhouse, which is time-consuming and labor-intensive, and helps to achieve a preliminary prediction of the time to market of agricultural cash crops within a region, thereby maximizing the economic benefits of agricultural cash crops.

Description

Technical Field

[0001] The invention relates to the technical field of agricultural economic crop growth period monitoring, and in particular to a crop growth period remote sensing monitoring method and electronic equipment in a plastic greenhouse. Background Art

[0002] Plastic greenhouses are an important agricultural management measure to ensure high yield and high quality of agricultural cash crops. The time when crops in the greenhouse are put on the market has a great impact on their sales price. Therefore, it is very necessary to accurately grasp the growth period of agricultural cash crops in regional greenhouses and then predict their time to market in order to maximize economic benefits. The traditional acquisition method is to use artificial ground surveys to investigate the growth period of agricultural cash crops, which not only consumes manpower and material resources, but also cannot effectively and accurately acquire data in a large area. In recent years, with the rapid development of remote sensing technology, a large-scale, fast and accurate monitoring method has been provided for the monitoring of plastic greenhouses. For specific agricultural cash crops that are widely planted on a large scale, the situation of agricultural cash crops in the greenhouse can be monitored by monitoring the spatial distribution of specific plastic greenhouses.

[0003] The main technical defect of existing remote sensing technology is that it mainly focuses on monitoring plastic greenhouses to monitor the distribution of specific agricultural economic crops. However, there is a lack of research on monitoring the growth period information of crops in greenhouses, and it still uses traditional manual survey methods, which is time-consuming and labor-intensive. Summary of the invention

[0004] The present invention provides a remote sensing monitoring method and electronic equipment for crop growth in a plastic greenhouse, which is used to solve the defects of the prior art that the growth period information monitoring of crops in the greenhouse uses the traditional manual investigation method, which is time-consuming and labor-intensive, and realizes the preliminary prediction of the time for agricultural cash crops to be put on the market within a regional range, thereby maximizing the economic benefits of agricultural cash crops.

[0005] The present invention provides a remote sensing monitoring method for crop growth in a plastic greenhouse, comprising:

[0006] Obtain remote sensing images of the plastic greenhouse where the crops are located;

[0007] Extracting object features based on the remote sensing image; the object features include spectral features, shape features and texture features;

[0008] Inputting the spectral features, the shape features and the texture features into a preset crop growth period classification model; wherein the crop growth period classification model is obtained by training the spectral features, the shape features and the texture features of different landforms;

[0009] The crop growth period conditions are judged based on the prediction results of the crop growth period classification model.

[0010] According to a remote sensing monitoring method for crop growth period in a plastic greenhouse provided by the present invention, the spectral characteristics include crop growth period parameters and conventional characteristics;

[0011] Wherein, the crop growth period parameter is obtained based on the reflectivity of the red band and the reflectivity of the green band of the remote sensing image;

[0012] The conventional features include a brightness value of the target object in the remote sensing image and a normalized vegetation parameter of the target object in the remote sensing image.

[0013] According to a remote sensing monitoring method for crop growth period in a plastic greenhouse provided by the present invention, the crop growth period parameters are obtained based on the following formula:

[0014]

[0015] Among them, NDPG represents the crop growth period parameter; R red and R green They represent the reflectance of the red band and the green band in the remote sensing image respectively.

[0016] According to a remote sensing monitoring method for crop growth period in a plastic greenhouse provided by the present invention, the crop growth period classification model is a decision tree model, and before the step of obtaining the remote sensing image of the plastic greenhouse where the crop is located, the step further includes:

[0017] Acquire sample data of different ground objects, select a first preset proportion of sample data from the sample data as a training set, and select a second preset proportion of sample data from the sample data as a validation set;

[0018] Setting the maximum depth of the decision tree, the maximum number of features, the spectral features, the shape features, and the texture features;

[0019] Starting from the root node, the training set is tested according to the preset features, and the tested training set is distributed to the child nodes according to the test results;

[0020] The training set is recursively redistributed for feature testing until a leaf node is reached, thereby generating a decision tree model in a recursive manner.

[0021] According to a remote sensing monitoring method for crop growth period in a plastic greenhouse provided by the present invention, the crop growth period classification model is a random forest model, and before the step of obtaining a remote sensing image of the plastic greenhouse where the crop is located, the method further includes:

[0022] Acquire sample data of different ground objects, select sample data of a third preset ratio among the sample data as a training set, and select sample data of a fourth preset ratio among the sample data as a validation set;

[0023] Setting the number of decision trees, the maximum depth, the maximum number of features, the spectral features, the shape features, and the texture features;

[0024] Randomly extract s samples with replacement from the training set, perform n samplings, and obtain n sample training sets;

[0025] A model is established based on each of the sampled training sets to obtain n decision tree models; the generated n decision trees constitute a random forest model.

[0026] According to a remote sensing monitoring method for crop growth in a plastic greenhouse provided by the present invention, after the step of obtaining remote sensing images of the plastic greenhouse where the crops are located, the method further includes:

[0027] A preprocessing operation is performed on the remote sensing image of the plastic greenhouse, wherein the preprocessing operation includes at least one of radiation correction, atmospheric correction, geometric correction and image fusion.

[0028] According to a remote sensing monitoring method for crop growth period in a plastic greenhouse provided by the present invention, after the step of preprocessing the remote sensing image of the plastic greenhouse and before the step of extracting object features based on the remote sensing image of the plastic greenhouse, the method further includes:

[0029] The remote sensing image is segmented based on at least one parameter of a segmentation scale, a shape parameter, and a compactness parameter.

[0030] The present invention also provides a remote sensing monitoring device for crop growth period in a plastic greenhouse, comprising:

[0031] The first acquisition module is used to acquire remote sensing images of the plastic greenhouse where the crops are located;

[0032] An object feature extraction module, used for extracting object features based on the remote sensing image; the object features include spectral features, shape features and texture features;

[0033] An input module, used to input the spectral features, the shape features and the texture features into a preset crop growth period classification model; wherein the crop growth period classification model is obtained by training the spectral features, the shape features and the texture features of different landforms;

[0034] The judgment module is used to judge the crop growth period situation based on the prediction result of the crop growth period classification model.

[0035] Further, the spectral characteristics include crop growth period parameters and conventional characteristics;

[0036] Wherein, the crop growth period parameter is obtained based on the reflectivity of the red band and the reflectivity of the green band of the remote sensing image;

[0037] The conventional features include a brightness value of the target object in the remote sensing image and a normalized vegetation parameter of the target object in the remote sensing image.

[0038] Furthermore, the crop growth period parameter is obtained based on the following formula:

[0039]

[0040] Among them, NDPG represents the crop growth period parameter; R red and R green They represent the reflectance of the red band and the green band in the remote sensing image respectively.

[0041] Furthermore, the remote sensing monitoring device for crop growth in a plastic greenhouse also includes:

[0042] A second acquisition module is used to acquire sample data of different ground objects, select sample data of a first preset ratio in the sample data as a training set, and select sample data of a second preset ratio in the sample data as a verification set;

[0043] A first setting module is used to set the maximum depth of the decision tree, the maximum number of features, the spectral feature, the shape feature and the texture feature;

[0044] The decision tree model generation module is used to start from the root node, test the training set according to preset features, and distribute the tested training set to child nodes according to the test results; recursively perform feature tests and redistribute the training set until a leaf node is reached, thereby generating a decision tree model in a recursive manner.

[0045] Furthermore, the remote sensing monitoring device for crop growth in a plastic greenhouse also includes:

[0046] A third acquisition module is used to acquire sample data of different ground objects, select sample data of a third preset ratio from the sample data as a training set, and select sample data of a fourth preset ratio from the sample data as a verification set;

[0047] A second setting module is used to set the number of decision trees, the maximum depth, the maximum number of features, the spectral features, the shape features, and the texture features;

[0048] The random forest model generation module is used to randomly extract s samples with replacement from the training set, and obtain n sampled training sets after n samplings; establish a model based on each of the sampled training sets to obtain n decision tree models; the generated n decision trees constitute a random forest model.

[0049] Furthermore, the remote sensing monitoring device for crop growth in a plastic greenhouse also includes:

[0050] The preprocessing operation module is used to perform preprocessing operations on the remote sensing image of the plastic greenhouse, and the preprocessing operations include at least one of radiation correction, atmospheric correction, geometric correction and image fusion.

[0051] Furthermore, the remote sensing monitoring device for crop growth in a plastic greenhouse also includes:

[0052] The image segmentation module is used to perform image segmentation on the remote sensing image, wherein the image segmentation is performed based on at least one parameter among a segmentation scale, a shape parameter, and a compactness parameter.

[0053] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of any of the above-mentioned methods for remote sensing monitoring of crop growth periods in a plastic greenhouse are implemented.

[0054] The present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the remote sensing monitoring method for crop growth period in a plastic greenhouse as described in any one of the above are implemented.

[0055] The present invention also provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps of any of the above-mentioned methods for remote sensing monitoring of crop growth periods in a plastic greenhouse.

[0056] The remote sensing monitoring method, device, storage medium and program product of the crop growth period in the plastic greenhouse provided by the present invention comprehensively considers the spectral characteristics, shape characteristics and texture characteristics of the plastic greenhouse where the crop is located, and inputs the spectral characteristics, shape characteristics and texture characteristics into a preset crop growth period classification model, wherein the crop growth period classification model is obtained by training the spectral characteristics, shape characteristics and texture characteristics of different landforms. Finally, the crop growth period is judged by the prediction result of the preset crop growth period classification model, thereby constructing a remote sensing monitoring method for the crop growth period (for example, early and mid-late period). Compared with the traditional time-consuming and laborious way of manually investigating the growth period of crops in plastic greenhouses, the embodiment of the present invention realizes remote judgment of the growth period of crops in plastic greenhouses through multiple feature information of remote sensing images of the plastic greenhouse where the crops are located, thereby realizing a preliminary prediction of the time to market of agricultural cash crops in plastic greenhouses, which is of great significance to maximizing the economic benefits of agricultural cash crops. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0058] Figure 1 This is one of the flow diagrams of the remote sensing monitoring method for crop growth period in a plastic greenhouse provided by the present invention;

[0059] Figure 2 A schematic diagram showing the comparison of spectral reflectance of early greenhouse Hami melon, mid-late greenhouse Hami melon, bare soil and vegetation in different bands in the prior art;

[0060] Figure 3 It is a structural schematic diagram of a crop growth period classification model provided by the present invention as a decision tree model;

[0061] Figure 4 It is a structural schematic diagram of a crop growth period classification model provided by the present invention as a random forest model;

[0062] Figure 5 This is the second flow chart of the crop growth period remote sensing monitoring method provided by the present invention;

[0063] Figure 6 This is the third flow chart of the crop growth period remote sensing monitoring method provided by the present invention;

[0064] Figure 7 It is a module schematic diagram of the remote sensing monitoring device for crop growth period in a plastic greenhouse provided by the present invention.

[0065] Figure 8 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. 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.

[0067] Combine the following Figure 1-Figure 6 The invention describes a remote sensing monitoring method for crop growth period in a plastic greenhouse.

[0068] Figure 1 FIG. 1 is a flow chart of a remote sensing monitoring method for crop growth in a plastic greenhouse according to an embodiment of the present invention. Figure 1 As shown, including:

[0069] S100, obtaining a remote sensing image of the plastic greenhouse where the crops are located;

[0070] The remote sensing image of the plastic greenhouse where the crops are located is obtained by electronic equipment. The plastic greenhouse is the growing area of ​​agricultural cash crops. Specifically, this step obtains cloud-free and high-resolution remote sensing image data of the target plastic greenhouse concentrated area by electronic equipment.

[0071] S400, extracting object features based on the remote sensing image; the object features include spectral features, shape features and texture features;

[0072] After the electronic device acquires the remote sensing image, object features are extracted based on the remote sensing image, including spectral features, shape features, and texture features.

[0073] Specifically, the spectral features include conventional features. Conventional features include the brightness value of the target object in the remote sensing image and the normalized vegetation parameters of the target object in the remote sensing image.

[0074] Among them, the brightness value of the target object in the remote sensing image and the

[0075] The normalized vegetation parameters of the target object are obtained by the following formula (1)-formula (3):

[0076] express:

[0077]

[0078]

[0079] NDVI object =(Mean nir -Mean red ) / (Mean nir +Mean red );

[0080] Formula (3)

[0081] Where n is the number of all pixels contained in the object in the remote sensing image; R n Refers to the reflectance value of the nth pixel; k is the number of bands of the remote sensing data used for research. In the embodiment of the present invention, k is 4, namely, blue, green, red and near infrared bands; Mean object Refers to the average reflectance of a certain object, where Mean nir 、Mean red 、Mean greenBrightness is the average reflectance of all pixels contained in an object in the near-infrared band, red band, and green band; object Refers to the brightness value of an object; NDVI object Refers to the normalized vegetation parameters of an object.

[0082] Regarding shape features, ten shape feature indicators are used in the embodiment of the present invention, including asymmetry, border index, compactness, density, elliptic fit, radius of largest enclosed ellipse, radius of smallest enclosing ellipse, rectangular fit, roundness, and shape index.

[0083] Regarding texture features, the gray-level co-occurrence matrix (GLCM) algorithm is used in the embodiment of the present invention to extract texture features of different objects. In some feasible embodiments, the texture features may include eight texture feature indicators, namely, homogeneity, contrast, dissimilarity, entropy, second moment, mean, standard deviation, and correlation.

[0084] S500, inputting the spectral features, the shape features, and the texture features into a preset crop growth period classification model; wherein the crop growth period classification model is obtained by training the spectral features, the shape features, and the texture features of different landforms;

[0085] The electronic device inputs the spectral features, shape features and texture features in the above steps into a preset crop growth period classification model. Based on the spectral features, shape features and texture features of the plastic greenhouse crops, the preset crop growth period classification model determines the growth period of the crops.

[0086] It should be noted that the preset crop growth period classification model is constructed before step S100. It is worth mentioning that the preset crop growth period classification model is obtained by training by comprehensively considering the object spectral characteristics, shape characteristics and texture characteristics of different ground objects.

[0087] S600: Determine the crop growth period situation based on the prediction result of the crop growth period classification model.

[0088] The crop growth period classification model classifies and determines the growth period of the crops in the plastic greenhouse corresponding to the remote sensing image based on the spectral characteristics, shape characteristics and texture characteristics of the crops in the plastic greenhouse.

[0089] Specifically, the standard for growth period classification can be set to two categories: samples of early growth and mid-to-late growth of greenhouse crops.

[0090] The crop growth period remote sensing monitoring method of this embodiment comprehensively considers the spectral characteristics, shape characteristics and texture characteristics of the plastic greenhouse where the crop is located, and inputs the spectral characteristics, shape characteristics and texture characteristics into a preset crop growth period classification model, wherein the crop growth period classification model is obtained by training the spectral characteristics, shape characteristics and texture characteristics of different landforms. Finally, the crop growth period is judged by the prediction results of the preset crop growth period classification model, thereby constructing a crop growth period (for example, early and mid-late) remote sensing monitoring method. Compared with the traditional time-consuming and labor-intensive method of manually investigating the growth period of crops in plastic greenhouses, the embodiment of the present invention uses multiple feature information of the remote sensing image of the plastic greenhouse where the crop is located to remotely judge the growth period of the crops in the plastic greenhouse, thereby realizing a preliminary prediction of the time to market of agricultural cash crops in the plastic greenhouse, which is of great significance to maximizing the economic benefits of agricultural cash crops.

[0091] Based on the above embodiments, as an optional embodiment, the spectral feature further includes a crop growth period parameter, wherein the crop growth period parameter is obtained based on the reflectivity of the red band and the reflectivity of the green band of the remote sensing image.

[0092] Through long-term observation by the applicant, it is found that the spectral characteristics of plastic greenhouses change as the crops in the greenhouse grow. In the early stage of crop growth in the greenhouse, the spectral reflectance characteristics of the plastic greenhouse are greatly affected by the soil, so they are similar to the spectral reflectance characteristics of bare soil; as the crops grow, the vegetation coverage gradually increases, and the spectral reflectance of the plastic greenhouse is gradually affected by the vegetation, and its reflection characteristics show reflection characteristics similar to those of vegetation. Figure 2 Take Hami melon grown in plastic greenhouses as an example. Figure 2 A schematic diagram showing the comparison of spectral reflectance of early greenhouse Hami melon, mid-late greenhouse Hami melon, bare soil and vegetation in different bands. Among them, b1, b2, b3 and b4 represent the blue band, green band, red band and near-infrared band respectively.

[0093] As the crops in the greenhouse grow, the reflectivity of the visible light band decreases, and the reflectivity of the near-infrared band increases significantly, gradually showing the spectral reflectivity characteristics of vegetation, especially in the green band, where the reflectivity is enhanced to form a reflection peak, while the red band absorption is gradually enhanced to form an absorption valley. Based on this phenomenon, this embodiment uses the red band reflectivity and green band reflectivity values ​​of remote sensing images to construct a crop growth period parameter, namely the normalized difference plastic greenhouse index (hereinafter referred to as NDPG). When the NDPG value is greater than or equal to 0, it is judged to be in the early stage of crop growth; when the NDPG value is negative, it is in the middle and late stages of crop growth, which has a good distinguishing effect on the identification of the growth period of crops in the greenhouse.

[0094] The crop growth period parameters are obtained based on the following formula (4):

[0095]

[0096] NDPG object =(Mean red -Mean green ) / (Mean red +Mean green ); formula (5)

[0097] Among them, NDPG represents the crop growth period parameter; R red and R green Respectively represent the reflectance of the red band and the reflectance of the green band in the remote sensing image. In formula (5), NDPG object Refers to the normalized plastic greenhouse index of a certain object.

[0098] By setting the normalized plastic greenhouse index as an important parameter for distinguishing the growth period of crops in the greenhouse, and inputting the spectral features with the normalized plastic greenhouse index into the crop growth period classification model, it is beneficial to further improve the classification accuracy of the crop growth period classification model in judging the crop growth period, especially for the sample recognition accuracy of the early growth of crops in the greenhouse, which is significantly improved, and effectively reduces the influence of bare soil on the judgment of the growth period of the crop.

[0099] Based on the above embodiments, as an optional embodiment, the crop growth period classification model is a decision tree model, and before the step of obtaining the remote sensing image of the plastic greenhouse where the crop is located, the step further includes:

[0100] S001. Obtain sample data of different ground objects, select sample data of a first preset ratio from the sample data as a training set, and select sample data of a second preset ratio from the sample data as a validation set;

[0101] Specifically, the different ground features in this embodiment include six categories: early-growing greenhouse Hami melons, mid- and late-growing greenhouse Hami melons, vegetation, water bodies, bare land, and others (such as building land, roads, beaches, and others). The first preset ratio and the second preset ratio can be set according to actual conditions, for example, 2 / 3 of the sample data are selected as the training set, and 1 / 3 of the sample data are selected as the verification set.

[0102] S002, setting the maximum depth of the decision tree, the maximum number of features, the spectral features, the shape features, and the texture features;

[0103] Specifically, the maximum depth of the decision tree can be set to 8 layers, and the maximum number of features can be set to 16. The spectral features select the conventional features and / or crop growth period parameters in the spectral features in step S400. The shape features select the ten shape feature indicators mentioned in step S400, and the texture features select the eight texture feature indicators mentioned in step S400.

[0104] S003, starting from the root node, testing the training set according to preset features, and distributing the tested training set to child nodes according to the test results;

[0105] S004. Recursively perform feature test redistribution on the training set until a leaf node is reached, and a decision tree model is generated in a recursive manner.

[0106] Specifically, starting from the root node, a feature of the training set is tested, and according to the test results, the tested training set is assigned to its child nodes. At this time, each child node corresponds to a value result of the feature. In this way, the training set is recursively tested and redistributed until it reaches the leaf node and is assigned to its class, and a decision tree is generated recursively. Finally, the prediction result is determined according to the judgment result of each node of the decision tree.

[0107] Please refer to the following Figure 3 and Figure 4 , Figure 3 It represents the decision tree model without the normalized plastic greenhouse index (i.e., crop growth period parameter); Figure 4 Represents a decision tree model with the normalized plastic greenhouse index. By considering the spectral characteristics, shape characteristics and texture characteristics of different objects, a remote sensing monitoring method for greenhouse crop growth period (early and mid-late period) based on an object-oriented decision tree algorithm was constructed to solve the time-consuming and laborious drawbacks of manual ground surveys during the growth period of traditional greenhouse economic crops, thereby achieving a preliminary prediction of the time to market for economic crops in the region, which is of great significance to maximizing the economic benefits of real crops.

[0108] Based on the above embodiments, as an optional embodiment, the crop growth period classification model is a random forest model, and before the step of obtaining the remote sensing image of the plastic greenhouse where the crop is located, the step further includes:

[0109] S005. Obtain sample data of different ground objects, select sample data of a third preset ratio from the sample data as a training set, and select sample data of a fourth preset ratio from the sample data as a validation set;

[0110] Similar to the decision tree model, the different ground features in this embodiment include six categories: early-growing greenhouse Hami melons, mid- and late-growing greenhouse Hami melons, vegetation, water bodies, bare land, and others (such as building land, roads, beaches, and others). 2 / 3 of the sample data are selected as the training set, and 1 / 3 of the sample data are selected as the validation set.

[0111] S006. Setting the number of decision trees, the maximum depth, the maximum number of features, the spectral features, the shape features, and the texture features;

[0112] Specifically, the number of decision trees can be set to 50, the maximum depth of the decision tree to 8 layers, and the maximum number of features to 16. The spectral features select the conventional features and / or crop growth period parameters in the spectral features in step S400. The shape features select the ten shape feature indicators mentioned in step S400, and the texture features select the eight texture feature indicators mentioned in step S400.

[0113] S007, randomly extracting s samples with replacement from the training set, performing n samplings, and obtaining n sample training sets;

[0114] S008. Establish a model based on each of the sampled training sets to obtain n decision tree models; the generated n decision trees constitute a random forest model.

[0115] Specifically, this embodiment uses the bootstrap method to randomly extract s samples with replacement from the sample data, and after n samplings, obtains n training sets; a model is established based on each new training set to obtain n decision tree models; the generated n decision trees are combined into a random forest, and the final prediction result is determined by voting of multiple tree classifiers.

[0116] The construction of an object-oriented random forest model is described below through an embodiment.

[0117] Taking the densely populated area of ​​Hami melons in greenhouses in Ledong Li Autonomous County, Hainan Province as the study area, the growth period monitoring research of Hami melons in plastic greenhouses was carried out using the GF-2 (spatial resolution 1 meter) images in January 2020.

[0118] The study obtained 1,336 ground feature sample data, including six categories: early-stage greenhouse Hami melon, mid- and late-stage greenhouse Hami melon, vegetation, water bodies, bare land, and others (construction land, roads, beaches, and others). 2 / 3 of the sample data were selected as the training set P_train, and the remaining 1 / 3 of the sample data were selected as the validation set T_test.

[0119] Set the number of decision trees to 50, the maximum feature to 16, and the other parameters to the system default. Select the spectral features involved in the classification (four band reflectance, NDVI, brightness value), 8 texture features (homogeneity, contrast, difference, entropy, second-order moment, mean, standard deviation, correlation) and 10 shape feature indicators (asymmetry, boundary index, compactness, density, ellipse fitting, radius of the largest enclosed ellipse, radius of the smallest enclosed ellipse, rectangle fitting, roundness, shape index).

[0120] Use the bootstrap method to randomly extract s samples with replacement from the sample data, and after n samplings, obtain n training sets; build models based on each new training set to obtain n decision tree models; form a random forest with the generated n decision trees, and use multiple tree classifiers to vote to decide the final prediction result.

[0121] The classification results of the decision tree model and the random forest model are described below through Table 1. PA is the producer accuracy (%); UA is the user accuracy (%); OA is the overall accuracy (%); KIA is the Kappa coefficient.

[0122]

[0123] Table 1

[0124] It can be seen from Table 1 that both the object-oriented decision tree model and the object-oriented random forest model have higher classification accuracy and have excellent beneficial effects compared with the traditional method of manually testing the crop growth period.

[0125] Further comparison of the decision tree model and the random forest model shows that the classification accuracy of the object-oriented random forest algorithm is better than that of the object-oriented decision tree algorithm; and the classification accuracy of the crop growth period classification model with the participation of NDPG (normalized plastic greenhouse index) is significantly higher than the classification accuracy without the participation of NDPG. Therefore, in some embodiments, the random forest model with the participation of NDPG (normalized plastic greenhouse index) is preferably used to predict the growth period of crops. The random forest model with the participation of NDPG (normalized plastic greenhouse index) significantly improves the sample recognition accuracy of the early growth of crops in the greenhouse, and effectively reduces the influence of bare soil on the crop samples.

[0126] Based on the above embodiments, as an optional embodiment, please refer to Figure 5 After the step of obtaining the remote sensing image of the plastic greenhouse where the crops are located, the method further includes:

[0127] S200, performing a preprocessing operation on the remote sensing image of the plastic greenhouse, wherein the preprocessing operation includes at least one of radiation correction, atmospheric correction, geometric correction and image fusion.

[0128] Since the original remote sensing images obtained have various interferences, such as the sensor itself, atmosphere, solar altitude angle, terrain, etc., radiation correction can eliminate the interference and obtain the real reflectivity data.

[0129] Atmospheric correction converts the radiation brightness or surface reflectivity into the actual reflectivity of the surface, with the aim of eliminating the errors caused by atmospheric scattering, absorption, and reflection. Atmospheric correction can be performed by using the absolute atmospheric correction method or the relative atmospheric correction method. The absolute atmospheric correction method can be performed by using the dark pixel method based on the simplified radiation transfer model, and the relative atmospheric correction method can be performed by using the histogram matching method.

[0130] Due to various factors such as shooting time, sensor posture, altitude, speed, etc., the objects in the image may be offset, stretched, deformed, etc. These geometric distortions need to be corrected to form a new image that meets certain map projection or graphic expression requirements.

[0131] Image fusion: Data from different sensors have different temporal, spatial and spectral resolutions and different polarization modes. The amount of image information obtained by a single sensor is limited and often cannot meet application needs. Image fusion can be used to obtain more useful information from different remote sensing images.

[0132] The above-mentioned preprocessing operations are helpful to eliminate various interferences and obtain more accurate remote sensing images, thereby improving the prediction accuracy of the crop growth period classification model, better judging the crop growth period, and achieving preliminary prediction of the time for agricultural cash crops to go on the market within the region, thereby maximizing the economic benefits of agricultural cash crops.

[0133] Based on the above embodiments, as an optional embodiment, please refer to Figure 6 After the step of preprocessing the remote sensing image of the plastic greenhouse and before the step of extracting object features based on the remote sensing image of the plastic greenhouse, the method further includes:

[0134] S300: performing image segmentation on the remote sensing image, wherein the image segmentation is performed based on at least one parameter of a segmentation scale, a shape parameter, and a compactness parameter.

[0135] The remote sensing image is segmented by an electronic device. Image segmentation is a key step in the object-oriented method, and a multi-scale segmentation algorithm is used to achieve accurate segmentation of different ground objects. Segmentation scale, shape parameter and compactness parameter are the main parameters that affect the segmentation effect of the multi-scale segmentation algorithm. In the specific segmentation process, it is necessary to try the settings of different factor parameters multiple times to obtain the best segmentation result. Specifically, in one embodiment, in the monitoring study of Hami melon growth period in a plastic greenhouse, the image segmentation parameter settings can be set to: segmentation scale 80, shape parameter 0.8, compactness parameter 0.9.

[0136] Please refer to Figure 7 , the remote sensing monitoring device for the growth period of crops in a plastic greenhouse provided by the present invention is described. The remote sensing monitoring device for the growth period of crops in a plastic greenhouse described below and the remote sensing monitoring method for the growth period of crops described above can be referenced to each other.

[0137] The first acquisition module 201 is used to acquire the remote sensing image of the plastic greenhouse where the crops are located;

[0138] An object feature extraction module 202 is used to extract object features based on the remote sensing image; the object features include spectral features, shape features and texture features;

[0139] An input module 203 is used to input the spectral features, the shape features and the texture features into a preset crop growth period classification model; wherein the crop growth period classification model is obtained by training the spectral features, the shape features and the texture features of different landforms;

[0140] The judgment module 204 is used to judge the crop growth period based on the prediction result of the crop growth period classification model.

[0141] The remote sensing monitoring device for the crop growth period in a plastic greenhouse provided by the embodiment of the present invention comprehensively considers the spectral characteristics, shape characteristics and texture characteristics of the plastic greenhouse where the crop is located, and inputs the spectral characteristics, shape characteristics and texture characteristics into a preset crop growth period classification model, wherein the crop growth period classification model is obtained by training the spectral characteristics, shape characteristics and texture characteristics of different landforms. Finally, the crop growth period is judged by the prediction result of the preset crop growth period classification model, thereby constructing a remote sensing monitoring method for the crop growth period (for example, early and mid-late period) in the greenhouse. Compared with the traditional time-consuming and laborious method of manually investigating the growth period of crops in plastic greenhouses, the embodiment of the present invention realizes remote judgment of the growth period of crops in plastic greenhouses through multiple feature information of remote sensing images of the plastic greenhouse where the crop is located, thereby realizing a preliminary prediction of the time to market of agricultural cash crops in plastic greenhouses, which is of great significance to maximizing the economic benefits of agricultural cash crops.

[0142] Based on the above embodiments, as an optional embodiment, the spectral characteristics include crop growth period parameters and conventional characteristics;

[0143] Wherein, the crop growth period parameter is obtained based on the reflectivity of the red band and the reflectivity of the green band of the remote sensing image;

[0144] The conventional features include a brightness value of the target object in the remote sensing image and a normalized vegetation parameter of the target object in the remote sensing image.

[0145] Based on the above embodiments, as an optional embodiment, the crop growth period parameter is obtained based on the following formula:

[0146]

[0147] Among them, NDPG represents the crop growth period parameter; R red and R green They represent the reflectance of the red band and the green band in the remote sensing image respectively.

[0148] Based on the above embodiments, as an optional embodiment, the remote sensing monitoring device for crop growth period in a plastic greenhouse further includes:

[0149] A second acquisition module is used to acquire sample data of different ground objects, select sample data of a first preset ratio in the sample data as a training set, and select sample data of a second preset ratio in the sample data as a verification set;

[0150] A first setting module is used to set the maximum depth of the decision tree, the maximum number of features, the spectral feature, the shape feature and the texture feature;

[0151] The decision tree model generation module is used to start from the root node, test the training set according to preset features, and distribute the tested training set to child nodes according to the test results; recursively perform feature tests and redistribute the training set until a leaf node is reached, thereby generating a decision tree model in a recursive manner.

[0152] Based on the above embodiments, as an optional embodiment, the remote sensing monitoring device for crop growth period in a plastic greenhouse further includes:

[0153] A third acquisition module is used to acquire sample data of different ground objects, select sample data of a third preset ratio from the sample data as a training set, and select sample data of a fourth preset ratio from the sample data as a verification set;

[0154] A second setting module is used to set the number of decision trees, the maximum depth, the maximum number of features, the spectral features, the shape features, and the texture features;

[0155] The random forest model generation module is used to randomly extract s samples with replacement from the training set, and obtain n sampled training sets after n samplings; establish a model based on each of the sampled training sets to obtain n decision tree models; the generated n decision trees constitute a random forest model.

[0156] Based on the above embodiments, as an optional embodiment, the remote sensing monitoring device for crop growth period in a plastic greenhouse further includes:

[0157] The preprocessing operation module is used to perform preprocessing operations on the remote sensing image of the plastic greenhouse, and the preprocessing operations include at least one of radiation correction, atmospheric correction, geometric correction and image fusion.

[0158] Based on the above embodiments, as an optional embodiment, the remote sensing monitoring device for crop growth period in a plastic greenhouse further includes:

[0159] The image segmentation module is used to perform image segmentation on the remote sensing image, wherein the image segmentation is performed based on at least one parameter among a segmentation scale, a shape parameter, and a compactness parameter.

[0160] Figure 8 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 8 As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830 and a communication bus 840, wherein the processor 810, the communication interface 820 and the memory 830 communicate with each other through the communication bus 840. The processor 810 may call the logic instructions in the memory 830 to execute the remote sensing monitoring method of the crop growth period in the plastic greenhouse, the method comprising: obtaining a remote sensing image of the plastic greenhouse where the crop is located; extracting object features based on the remote sensing image; the object features include spectral features, shape features and texture features; inputting the spectral features, the shape features and the texture features into a preset crop growth period classification model; wherein the crop growth period classification model is obtained by training the spectral features, the shape features and the texture features of different landforms; judging the crop growth period based on the prediction results of the crop growth period classification model.

[0161] In addition, the logic instructions in the above-mentioned memory 830 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, 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, and the computer software product is stored in a storage medium, including a number of 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 method 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 and other media that can store program codes.

[0162] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the remote sensing monitoring method of crop growth period in a plastic greenhouse provided by the above-mentioned methods, and the method includes: obtaining a remote sensing image of the plastic greenhouse where the crops are located; extracting object features based on the remote sensing image; the object features include spectral features, shape features and texture features; inputting the spectral features, the shape features and the texture features into a preset crop growth period classification model; wherein the crop growth period classification model is obtained by training the spectral features, the shape features and the texture features of different landforms; and judging the crop growth period situation based on the prediction results of the crop growth period classification model.

[0163] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the remote sensing monitoring method for crop growth period in a plastic greenhouse provided by the above-mentioned methods, the method comprising: obtaining a remote sensing image of the plastic greenhouse where the crops are located; extracting object features based on the remote sensing image; the object features include spectral features, shape features and texture features; inputting the spectral features, the shape features and the texture features into a preset crop growth period classification model; wherein the crop growth period classification model is obtained by training the spectral features, the shape features and the texture features of different landforms; and judging the crop growth period situation based on the prediction results of the crop growth period classification model.

[0164] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0165] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0166] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A remote sensing monitoring method for crop growth in a plastic greenhouse, characterized in that: include: Obtain remote sensing images of the plastic greenhouse where the crops are located; Extracting object features based on the remote sensing image; the object features include spectral features, shape features and texture features; Inputting the spectral features, the shape features and the texture features into a preset crop growth period classification model; wherein the crop growth period classification model is obtained by training the spectral features, the shape features and the texture features of different landforms; Determining the crop growth period based on the prediction results of the crop growth period classification model; The spectral characteristics include crop growth period parameters and conventional characteristics; Wherein, the crop growth period parameter is obtained based on the reflectivity of the red band and the reflectivity of the green band of the remote sensing image; The conventional features include a brightness value of the target object in the remote sensing image and a normalized vegetation parameter of the target object in the remote sensing image; The crop growth period parameters are obtained based on the following formula: ； Among them, NDPG represents crop growth period parameters; and They represent the reflectance of the red band and the green band in the remote sensing image respectively.

2. The remote sensing monitoring method for crop growth in a plastic greenhouse according to claim 1, characterized in that: The crop growth period classification model is a decision tree model, and the step of obtaining the remote sensing image of the plastic greenhouse where the crop is located also includes: Acquire sample data of different ground objects, select a first preset proportion of sample data from the sample data as a training set, and select a second preset proportion of sample data from the sample data as a validation set; Setting the maximum depth of the decision tree, the maximum number of features, the spectral features, the shape features, and the texture features; Starting from the root node, the training set is tested according to the preset features, and the tested training set is distributed to the child nodes according to the test results; The training set is recursively redistributed for feature testing until a leaf node is reached, thereby generating a decision tree model in a recursive manner.

3. The remote sensing monitoring method for crop growth in a plastic greenhouse according to claim 1, characterized in that: The crop growth period classification model is a random forest model, and the step of obtaining the remote sensing image of the plastic greenhouse where the crop is located also includes: Acquire sample data of different ground objects, select sample data of a third preset ratio from the sample data as a training set, and select sample data of a fourth preset ratio from the sample data as a validation set; Setting the number of decision trees, the maximum depth, the maximum number of features, the spectral features, the shape features, and the texture features; Randomly extract s samples with replacement from the training set, perform n samplings, and obtain n sample training sets; A model is established based on each of the sampled training sets to obtain n decision tree models; the generated n decision trees constitute a random forest model.

4. The remote sensing monitoring method for crop growth in a plastic greenhouse according to claim 1, characterized in that: After the step of obtaining the remote sensing image of the plastic greenhouse where the crops are located, the method further includes: A preprocessing operation is performed on the remote sensing image of the plastic greenhouse, wherein the preprocessing operation includes at least one of radiation correction, atmospheric correction, geometric correction and image fusion.

5. The remote sensing monitoring method for crop growth in a plastic greenhouse according to claim 4, characterized in that: After the step of preprocessing the remote sensing image of the plastic greenhouse and before the step of extracting object features based on the remote sensing image of the plastic greenhouse, the method further includes: The remote sensing image is segmented based on at least one parameter of a segmentation scale, a shape parameter, and a compactness parameter.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the remote sensing monitoring method for crop growth period in a plastic greenhouse as described in any one of claims 1 to 5 are implemented.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the remote sensing monitoring method for crop growth period in a plastic greenhouse as claimed in any one of claims 1 to 5 are implemented.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the remote sensing monitoring method for crop growth period in a plastic greenhouse as claimed in any one of claims 1 to 5 are implemented.

Images