CN117292282A - Afforestation growth condition monitoring method and system based on high-resolution unmanned aerial vehicle remote sensing - Google Patents
Afforestation growth condition monitoring method and system based on high-resolution unmanned aerial vehicle remote sensing Download PDFInfo
- Publication number
- CN117292282A CN117292282A CN202311485354.5A CN202311485354A CN117292282A CN 117292282 A CN117292282 A CN 117292282A CN 202311485354 A CN202311485354 A CN 202311485354A CN 117292282 A CN117292282 A CN 117292282A
- Authority
- CN
- China
- Prior art keywords
- landscaping
- data
- growth
- module
- color
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000012010 growth Effects 0.000 title claims abstract description 124
- 238000000034 method Methods 0.000 title claims abstract description 43
- 238000012544 monitoring process Methods 0.000 title claims abstract description 35
- 238000004458 analytical method Methods 0.000 claims abstract description 57
- 238000011835 investigation Methods 0.000 claims abstract description 45
- 238000012545 processing Methods 0.000 claims abstract description 20
- 238000002372 labelling Methods 0.000 claims abstract description 7
- 238000004364 calculation method Methods 0.000 claims description 32
- 238000012549 training Methods 0.000 claims description 29
- 239000003086 colorant Substances 0.000 claims description 24
- 238000012360 testing method Methods 0.000 claims description 22
- 239000002689 soil Substances 0.000 claims description 17
- 244000025254 Cannabis sativa Species 0.000 claims description 13
- 230000000877 morphologic effect Effects 0.000 claims description 12
- 238000011156 evaluation Methods 0.000 claims description 11
- 238000012937 correction Methods 0.000 claims description 6
- 230000000007 visual effect Effects 0.000 claims description 6
- 230000007774 longterm Effects 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 230000008635 plant growth Effects 0.000 claims description 3
- 230000000306 recurrent effect Effects 0.000 claims description 3
- 230000006403 short-term memory Effects 0.000 claims description 3
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 3
- 230000015654 memory Effects 0.000 claims description 2
- 238000012800 visualization Methods 0.000 claims 1
- 238000010276 construction Methods 0.000 abstract description 4
- 238000002474 experimental method Methods 0.000 description 6
- 238000002790 cross-validation Methods 0.000 description 4
- 238000005457 optimization Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 241000196324 Embryophyta Species 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000010200 validation analysis Methods 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/17—Terrestrial scenes taken from planes or by drones
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/24—Aligning, centring, orientation detection or correction of the image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/30—Noise filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/34—Smoothing or thinning of the pattern; Morphological operations; Skeletonisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/188—Vegetation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Remote Sensing (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a landscaping growth monitoring method and system based on high-resolution unmanned aerial vehicle remote sensing, comprising a basic image data module, an image data processing module, a landscaping object labeling data module, an outside industry investigation data module, a landscaping object characteristic data module, a data set module and a landscaping growth analysis module, wherein the outside industry investigation data module and the data output by the landscaping object characteristic data module generate the data set module, and the construction of a growth analysis model is carried out according to a large number of samples based on the principle that the landscaping objects under different growth conditions are different in expression forms on unmanned aerial vehicle images, so that a convenient and efficient landscaping growth analysis method is realized.
Description
Technical Field
The invention relates to the technical field of garden management, in particular to a garden greening growth condition monitoring method and system based on high-resolution unmanned aerial vehicle remote sensing.
Background
With the acceleration of the urban process in recent years, urban landscaping construction plays an increasingly important role in urban planning and construction. The fund investment of urban landscaping is continuously increased, and the area of urban landscaping is continuously enlarged. However, the landscaping growth monitoring technology still has certain hysteresis.
The traditional landscaping growth condition monitoring method mainly relies on manual field collection and measurement and has the following defects:
(1) The collection efficiency is low: traditional afforestation growth monitoring generally relies on manual visual observation and measurement, requires a large amount of labor investment and time cost, is low in large-area manual monitoring efficiency, and limits the scale and frequency of monitoring.
(2) The cost is high: traditional landscaping growth monitoring also requires manual data collection, processing and analysis, and the use of specialized equipment. These costs include labor costs, equipment costs, and time costs, making the monitoring costly.
(3) Subjectivity and inconsistency: because of the influence of human factors, manual monitoring is easily influenced by subjective judgment and individual differences, resulting in inconsistency of monitoring results. Different observers may have different evaluation criteria and judgment criteria, thereby affecting the accuracy and comparability of the monitoring results.
(4) Timeliness and timeliness: traditional landscaping growth monitoring is usually carried out regularly, and vegetation change conditions cannot be obtained in real time. Such delays may lead to timely intervention for vegetation health and inadequate decision support.
In view of the above, the invention provides a high-resolution unmanned aerial vehicle remote sensing-based landscaping growth condition monitoring method and system, which are based on the principle that landscaping objects under different growth conditions have different expression forms on unmanned aerial vehicle images, and the construction of a growth condition analysis model is carried out according to a large number of samples, so that a convenient and efficient landscaping growth condition analysis method is realized.
Disclosure of Invention
The invention aims to provide a high-resolution unmanned aerial vehicle remote sensing-based landscaping growth condition monitoring method and system, which are based on the principle that landscaping objects under different growth conditions have different expression forms on unmanned aerial vehicle images, and a growth condition analysis model is constructed according to a large number of samples, so that a convenient and efficient landscaping growth condition analysis method is realized.
A landscaping growth monitoring method based on high-resolution unmanned aerial vehicle remote sensing is characterized by comprising the following steps:
step1, acquiring basic image data, and acquiring remote sensing images in the urban landscaping range by using an unmanned aerial vehicle mounting camera.
Step2: and processing the basic image data, and carrying out orthographic correction, image stitching, noise processing and image enhancement on the obtained basic remote sensing image data.
Step3: afforestation object annotation data and field investigation data are obtained, the landscaping object comprises trees, shrubs and ground cover, the landscaping object annotation data comprise landscaping object categories and geographic position information, the field investigation data comprise tree height information, tree crown form information, blade color information, blade area index information, grassland coverage information and soil condition information, the landscaping objects in the processed image data are subjected to frame selection annotation, the categories and the geographic position information of the landscaping objects are recorded, and the field investigation is carried out on the landscaping objects based on the geographic position information to obtain the field investigation data.
Step4, obtaining landscaping object feature data based on landscaping object marking data and field investigation data, wherein the landscaping object feature data comprises morphological features, texture features, color features and spatial relationship features, the morphological features comprise tree crown shapes, trunk diameters and tree heights, the texture features comprise leaf textures, shrub textures and grassland textures, the color features comprise leaf colors, shrub colors and grassland colors, and the spatial relationship features comprise tree relative positions and distribution rules.
And Step5, integrating and forming a data set based on the characteristic data of the landscaping objects and the field investigation data, wherein the data set comprises a training set and a testing set, and constructing a landscaping growth analysis model based on the training set.
And Step6, evaluating the landscaping growth analysis model by using a test set, wherein evaluation data comprise calculation accuracy, recall rate indexes and prediction results of the visual model.
In some embodiments, step3, further comprising the Step of field survey data including tree height information, crown shape information, leaf color information, leaf area index information, lawn coverage information, soil condition information,
the tree height information is the height of the tree, higher trees tend to represent better growth conditions, the crown morphology information is the crown morphology of the tree, larger, dense and symmetrical crowns generally represent better growth conditions,
the height, crown size and crown density of the tree are judged according to the average statistical value of other healthy trees of the same variety and the same age, if the average statistical value is higher than the average statistical value, the growth condition is considered to be better, and if the average statistical value is lower than the average statistical value, the artificial growth condition is poorer.
The color information of the leaves is the color of the tree leaves, the color of the tree leaves is usually light green or bright green, if the leaves are yellow or withered, the leaves may represent the problems of poor growth or water shortage, etc.,
leaf Area Index (LAI) information is a ratio of the coverage Area of tree leaves to the ground Area, and is used for evaluating the degree of complexity and growth condition of tree leaves, a higher LAI value generally indicates a better growth condition, a general LAI value ranges from 1 to 10, a larger value indicates a better growth condition, but the relation between a specific value and the plant growth condition cannot be roughly said, plant varieties are required to be matched for discrimination, and the best discrimination standard is that the average value is higher than the average value of LAI of healthy vegetation of the same variety, and the average value is artificially better.
The grass coverage information is the coverage of the grass for assessing the growth status, a higher coverage generally indicates that the grass is growing well, a value of the grass coverage is between 0 and 1, 0 indicates worst, 1 indicates best, and the soil condition information is the soil condition for judging the growth status, and in a better grass growth status, the soil should be covered by effective vegetation without bare soil exposure.
In some embodiments, step4 further includes the Step of calculating the shape of the crown in the landscaping object feature data in such a way that the Crown Width (CW) refers to the maximum extension of the crown, and generally measuring the east-west width (EW) and the north-south width (NS) of the crown, where the calculation formula is: CW= (EW+NS)/2, the Crown Area (CA) refers to the coverage area of the crown vertical projection ground, the texture characteristic index and the calculation formula.
Further, the texture feature in the landscaping object feature data is calculated by Entropy (Entropy): to describe the complexity and texture richness of the image. The calculation formula is as follows: entropy= Σ (P (i) ×log2 (P (i))), where P (i) is the probability of gray level per pel,
contrast (Contrast): to describe the degree of gray scale difference between adjacent pixels in an image. The calculation formula is as follows: contrast= Σ ((i- μ) ≡2×P (i)), where i is the gray level of each pixel, μ is the average gray level of the image, energy (Energy): describing the texture details of the image, and calculating the following formula: energy= Σ (P (i)/(2), where P (i) is the probability of each pixel, correlation): to describe the correlation between the gray level of a picture element and neighboring picture elements in an image. The calculation formula is as follows: corelation= Σ ((i- μ) (j- μ) P (i, j)/(σ_i×σ_j), where i and j are gray levels of adjacent pixels, P (i, j) is a joint probability of adjacent pixels, μ is an average gray level of an image, σ is a standard deviation of gray levels, uniformity (Uniformity): the method is used for describing the consistency between pixel gray levels in the image, and the calculation formula is as follows: uniformity= Σ (P (i)/(2), where P (i) is the gray level probability of each pixel, degree of freedom (Degree of Freedom): the degree of gray level variation near the picture elements in the image is described. The calculation formula is as follows: degree of Freedom = Σ ((i-j)/(2×p (i, j))/(P (i, j)), where i and j are the gray levels of neighboring pixels and P (i, j) is the joint probability of neighboring pixels.
Further, the Color characteristic index and the calculation mode in the landscaping object characteristic data are Average colors (Average colors): color features are described by calculating average color values for pixels within an object region. An average value in a Color space such as RGB, HSV, lab and a Color Contrast (Color Contrast) can be calculated: to describe the degree of difference between the object region color and the background color in the image. The difference between the Color histogram of the object region and the background Color histogram, color Uniformity (Color Uniformity), may be calculated: to describe uniformity of color and uniformity of distribution of image areas. The variance or standard deviation of the color histogram may be calculated to obtain the color uniformity index.
Further, the calculating mode of the spatial relation characteristic index in the landscaping object characteristic data comprises the following steps: the tree relative position refers to the distance and the azimuth of the current target object relative to other objects, and the distribution rule refers to the arrangement and combination rule of the objects of the same category in a certain space range.
In some embodiments, in Step5, the method further comprises the Step of training the landscaping growth analysis model with the training set and evaluating the performance of the landscaping growth analysis model with the testing set.
Further, the training set accounts for 70% and the test set accounts for 30%, K-fold cross validation is adopted for 70% of the training set, when K-fold cross validation is adopted, 70% of training set data can be equally divided into K equal parts, K independent learning experiments are then carried out, one part is selected as validation in each experiment, the rest K-1 parts are used as training, performance evaluation of a model is prevented from being influenced by accidental errors caused by single division, a Bayesian optimization model is adopted by adopting a super-parameter optimization strategy in each experiment, and an optimal landscaping growth analysis model is obtained.
Further, the landscaping growth analysis model adopts a convolution long-short-Term Memory recurrent neural network model (Convolutional Long Short-Term Memory, convLSTM), the landscaping growth analysis model comprises a network model input layer, a convolution layer, a hiding layer and an output layer, the input layer is used for receiving basic remote sensing image data and field investigation data of each period and converting the basic remote sensing image data and the field investigation data into a data type which can be identified by the landscaping growth analysis model, the convolution layer is used for extracting features from the basic remote sensing image data, the hiding layer is composed of a plurality of ConvLSTM units, f in each ConvLSTM unit represents a forgetting gate, the forgetting gate determines what information should be forgotten, i represents an input gate, the input gate determines how much new information can be added into the state, o represents an output gate, the output gate controls the state of cells to be output to the next layer, c represents the state of cells, h represents the hiding state, and is used for outputting and transferring the hidden state to ConvLSTM units of the next time step, t represents a time state, and represents a batched LSTM of input data, the ConvLSTM unit passes through the forgetting gate, the forgetting gate and the input gate can memorize the input state and the long-Term depending on the input program; and the output layer of the final model converts the output data of the hidden layer into a growth analysis result.
Further, the landscaping growth analysis model performs performance evaluation based on 30% of the test set, the landscaping growth analysis model is applied to other areas to perform landscaping growth analysis, remote sensing images acquired by high-resolution unmanned aerial vehicles in other areas are input to the landscaping growth analysis model, after the calculation of the landscaping growth analysis model is completed, the growth results of each landscaping element in other areas are output, and the growth results are displayed in a two-dimensional visual mode.
A system for monitoring the growth of landscaping based on high-resolution unmanned aerial vehicle remote sensing comprises a basic image data module, an image data processing module, a landscaping object labeling data module, an outside industry investigation data module, a landscaping object characteristic data module, a data set module and a landscaping growth analysis module, wherein the outside industry investigation data module and the data output by the landscaping object characteristic data module generate a data set module, the data set module comprises a training set module and a test set module,
and a basic image data module: the remote sensing image acquiring method is used for acquiring a remote sensing image in the urban landscaping range by using the unmanned aerial vehicle mounting camera;
the image data processing module: the method is used for processing the basic image data, and carrying out orthographic correction, image stitching, noise processing and image enhancement on the acquired basic remote sensing image data;
afforestation object label data module: the method comprises the steps of performing frame selection labeling on landscaping objects in processed image data, and recording category and geographic position information of the landscaping objects;
field investigation data module: the method comprises the steps of performing field investigation on landscaping objects based on geographic position information to obtain field investigation data;
the landscaping object feature data module is used for obtaining landscaping object feature data based on output data of the landscaping object annotation data module and output data of the field investigation data module, the landscaping object feature data comprises morphological features, texture features, color features and spatial relationship features, the morphological features comprise crown shapes, trunk diameters and tree heights, the texture features comprise leaf textures, shrub textures and grassland textures, the color features comprise leaf colors, shrub colors and grassland colors, and the spatial relationship features comprise tree relative positions and distribution rules;
the data set module comprises a training set module and a testing set module, wherein the training set module is used for generating an afforestation growth analysis module, and the testing set module is used for evaluating the afforestation growth analysis module and outputting evaluation data.
The invention has the beneficial effects that: based on the principle that landscaping objects under different growth conditions have different expression forms on unmanned aerial vehicle images, a growth analysis model is constructed according to a large number of samples, and a convenient and efficient landscaping growth analysis method is realized.
Drawings
The foregoing and other features of the present application will be more fully described when read in conjunction with the following drawings. It is appreciated that these drawings depict only several embodiments of the present application and are therefore not to be considered limiting of its scope. The present application will be described more specifically and in detail by using the accompanying drawings.
Fig. 1 is a schematic flow chart of a method and a system for monitoring landscaping growth based on high-resolution unmanned aerial vehicle remote sensing.
Fig. 2 is a schematic diagram of a single ConvLSTM unit structure of the method and system for monitoring landscaping growth based on high-resolution unmanned aerial vehicle remote sensing according to the present invention.
Detailed Description
The following examples are described to aid in the understanding of the present application and are not, nor should they be construed in any way to limit the scope of the present application.
In the following description, those skilled in the art will recognize that components may be described as separate functional units (which may include sub-units) throughout this discussion, but those skilled in the art will recognize that various components or portions thereof may be divided into separate components or may be integrated together (including integration within a single system or component).
Meanwhile, the connection between components or systems is not intended to be limited to a direct connection. Rather, data between these components may be modified, reformatted, or otherwise changed by intermediate components. In addition, additional or fewer connections may be used. It should also be noted that the terms "coupled," "connected," or "input" are to be construed as including direct connection, indirect connection or fixation through one or more intermediaries.
Example 1:
as shown in fig. 1, the flow diagram of the landscaping growth monitoring method and system based on the high-resolution unmanned aerial vehicle remote sensing according to the invention is shown; fig. 2 is a schematic diagram of a single ConvLSTM unit structure of the method and system for monitoring landscaping growth based on high-resolution unmanned aerial vehicle remote sensing according to the present invention.
A landscaping growth monitoring method based on high-resolution unmanned aerial vehicle remote sensing is characterized by comprising the following steps:
step1, acquiring basic image data, and acquiring remote sensing images in the urban landscaping range by using an unmanned aerial vehicle mounting camera.
Step2: and processing the basic image data, and carrying out orthographic correction, image stitching, noise processing and image enhancement on the obtained basic remote sensing image data.
Step3: afforestation object annotation data and field investigation data are obtained, the landscaping object comprises trees, shrubs and ground cover, the landscaping object annotation data comprise landscaping object categories and geographic position information, the field investigation data comprise tree height information, tree crown form information, blade color information, blade area index information, grassland coverage information and soil condition information, the landscaping objects in the processed image data are subjected to frame selection annotation, the categories and the geographic position information of the landscaping objects are recorded, and the field investigation is carried out on the landscaping objects based on the geographic position information to obtain the field investigation data.
Step4, obtaining landscaping object feature data based on landscaping object marking data and field investigation data, wherein the landscaping object feature data comprises morphological features, texture features, color features and spatial relationship features, the morphological features comprise tree crown shapes, trunk diameters and tree heights, the texture features comprise leaf textures, shrub textures and grassland textures, the color features comprise leaf colors, shrub colors and grassland colors, and the spatial relationship features comprise tree relative positions and distribution rules.
And Step5, integrating and forming a data set based on the characteristic data of the landscaping objects and the field investigation data, wherein the data set comprises a training set and a testing set, and constructing a landscaping growth analysis model based on the training set.
And Step6, evaluating the landscaping growth analysis model by using a test set, wherein evaluation data comprise calculation accuracy, recall rate indexes and prediction results of the visual model.
In Step3, the method further comprises the Step of carrying out field investigation data including tree height information, crown shape information, leaf color information, leaf area index information, grassland coverage information and soil condition information,
the tree height information is the height of the tree, higher trees tend to represent better growth conditions, the crown morphology information is the crown morphology of the tree, larger, dense and symmetrical crowns generally represent better growth conditions,
the height, crown size and crown density of the tree are judged according to the average statistical value of other healthy trees of the same variety and the same age, if the average statistical value is higher than the average statistical value, the growth condition is considered to be better, and if the average statistical value is lower than the average statistical value, the artificial growth condition is poorer.
The color information of the leaves is the color of the tree leaves, the color of the tree leaves is usually light green or bright green, if the leaves are yellow or withered, the leaves may represent the problems of poor growth or water shortage, etc.,
leaf Area Index (LAI) information is a ratio of the coverage Area of tree leaves to the ground Area, and is used for evaluating the degree of complexity and growth condition of tree leaves, a higher LAI value generally indicates a better growth condition, a general LAI value ranges from 1 to 10, a larger value indicates a better growth condition, but the relation between a specific value and the plant growth condition cannot be roughly said, plant varieties are required to be matched for discrimination, and the best discrimination standard is that the average value is higher than the average value of LAI of healthy vegetation of the same variety, and the average value is artificially better.
The grass coverage information is the coverage of the grass for assessing the growth status, a higher coverage generally indicates that the grass is growing well, a value of the grass coverage is between 0 and 1, 0 indicates worst, 1 indicates best, and the soil condition information is the soil condition for judging the growth status, and in a better grass growth status, the soil should be covered by effective vegetation without bare soil exposure.
In Step4, the method further includes the following steps, wherein the crown shape in the landscaping object feature data is calculated in such a way that the Crown Width (CW) refers to the maximum extension of the crown, and the east-west width (EW) and the north-south width (NS) of the crown are generally measured, and the calculation formula is as follows: CW= (EW+NS)/2, the Crown Area (CA) refers to the coverage area of the crown vertical projection ground, the texture characteristic index and the calculation formula.
The texture feature calculation mode in the landscaping object feature data is Entropy (Entropy): to describe the complexity and texture richness of the image. The calculation formula is as follows: entropy= Σ (P (i) x)
log2 (P (i)), where P (i) is the probability of gray level per pixel,
contrast (Contrast): to describe the degree of gray scale difference between adjacent pixels in an image. The calculation formula is as follows: contrast= Σ ((i- μ) ≡2×P (i)), where i is the gray level of each pixel, μ is the average gray level of the image, energy (Energy): describing the texture details of the image, and calculating the following formula: energy= Σ (P (i)/(2), where P (i) is the probability of each pixel, correlation): to describe the correlation between the gray level of a picture element and neighboring picture elements in an image. The calculation formula is as follows: corelation= Σ ((i- μ) (j- μ) P (i, j)/(σ_i×σ_j), where i and j are gray levels of adjacent pixels, P (i, j) is a joint probability of adjacent pixels, μ is an average gray level of an image, σ is a standard deviation of gray levels, uniformity (Uniformity): the method is used for describing the consistency between pixel gray levels in the image, and the calculation formula is as follows: uniformity= Σ (P (i)/(2), where P (i) is the gray level probability of each pixel, degree of freedom (Degree of Freedom): the degree of gray level variation near the picture elements in the image is described. The calculation formula is as follows: degree of Freedom = Σ ((i-j)/(2×p (i, j))/(P (i, j)), where i and j are the gray levels of neighboring pixels and P (i, j) is the joint probability of neighboring pixels.
The Color characteristic index and the calculation mode in the landscaping object characteristic data are Average colors (Average colors): color features are described by calculating average color values for pixels within an object region. An average value in a Color space such as RGB, HSV, lab and a Color Contrast (Color Contrast) can be calculated: to describe the degree of difference between the object region color and the background color in the image. The difference between the Color histogram of the object region and the background Color histogram, color Uniformity (Color Uniformity), may be calculated: to describe uniformity of color and uniformity of distribution of image areas. The variance or standard deviation of the color histogram may be calculated to obtain the color uniformity index.
The calculating mode of the spatial relation characteristic index in the landscaping object characteristic data comprises the following steps: the tree relative position refers to the distance and the azimuth of the current target object relative to other objects, and the distribution rule refers to the arrangement and combination rule of the objects of the same category in a certain space range.
In Step5, the method further comprises the following steps that the training set is used for training the landscaping growth analysis model, and the test set is used for evaluating performance of the landscaping growth analysis model.
According to the training set accounting for 70% and the testing set accounting for 30%, K-fold cross validation is adopted for 70% of the training sets, when K-fold cross validation is adopted, 70% of training set data can be equally divided into K equal parts, K independent learning experiments are then carried out, one part is selected as validation in each experiment, the rest K-1 parts are used as training, performance evaluation of a model is prevented from being influenced by accidental errors caused by single division, a Bayesian optimization model is adopted by adopting a super-parameter optimization strategy in each experiment, and an optimal landscaping growth condition analysis model is obtained.
The landscaping growth analysis model adopts a convolution long-Term Memory recurrent neural network model (Convolutional Long Short-Term Memory, convLSTM), the landscaping growth analysis model comprises a network model input layer, a convolution layer, a hiding layer and an output layer, the input layer is used for receiving basic remote sensing image data and field investigation data in each period and converting the basic remote sensing image data and the field investigation data into a data type which can be identified by the landscaping growth analysis model, the convolution layer is used for extracting features from the basic remote sensing image data, the hiding layer is composed of a plurality of ConvLSTM units, f in each ConvLSTM unit represents a forgetting gate, the forgetting gate determines what information should be forgotten, i represents an input gate, the input gate determines which new information can be added into the state, o represents an output gate, the output gate controls how many cell states are output to the next layer, h represents a hidden state and is used for outputting and transmitting the ConvLSTM unit of the next time step, t represents a time state, BN represents a batch standardization program for the input data, the ConLSTM unit passes through the forgetting gate, the input gate and the input state can learn the data and the long-Term dependence on the input state; and the output layer of the final model converts the output data of the hidden layer into a growth analysis result.
The landscaping growth analysis model carries out performance evaluation based on 30% of test sets, the landscaping growth analysis model is applied to other areas to carry out landscaping growth analysis, remote sensing images acquired by high-resolution unmanned aerial vehicles in other areas are input into the landscaping growth analysis model, after the calculation of the landscaping growth analysis model is completed, the growth result of each landscaping element in other areas is output, and the growth result is subjected to two-dimensional visual presentation.
A system for monitoring the growth of landscaping based on high-resolution unmanned aerial vehicle remote sensing comprises a basic image data module, an image data processing module, a landscaping object labeling data module, an outside industry investigation data module, a landscaping object characteristic data module, a data set module and a landscaping growth analysis module, wherein the outside industry investigation data module and the data output by the landscaping object characteristic data module generate a data set module, the data set module comprises a training set module and a test set module,
and a basic image data module: the remote sensing image acquiring method is used for acquiring a remote sensing image in the urban landscaping range by using the unmanned aerial vehicle mounting camera;
the image data processing module: the method is used for processing the basic image data, and carrying out orthographic correction, image stitching, noise processing and image enhancement on the acquired basic remote sensing image data;
afforestation object label data module: the method comprises the steps of performing frame selection labeling on landscaping objects in processed image data, and recording category and geographic position information of the landscaping objects;
field investigation data module: the method comprises the steps of performing field investigation on landscaping objects based on geographic position information to obtain field investigation data;
the landscaping object feature data module is used for obtaining landscaping object feature data based on output data of the landscaping object annotation data module and output data of the field investigation data module, the landscaping object feature data comprises morphological features, texture features, color features and spatial relationship features, the morphological features comprise crown shapes, trunk diameters and tree heights, the texture features comprise leaf textures, shrub textures and grassland textures, the color features comprise leaf colors, shrub colors and grassland colors, and the spatial relationship features comprise tree relative positions and distribution rules;
the data set module comprises a training set module and a testing set module, wherein the training set module is used for generating an afforestation growth analysis module, and the testing set module is used for evaluating the afforestation growth analysis module and outputting evaluation data.
The invention has the beneficial effects that: based on the principle that landscaping objects under different growth conditions have different expression forms on unmanned aerial vehicle images, a growth analysis model is constructed according to a large number of samples, and a convenient and efficient landscaping growth analysis method is realized.
While various aspects and embodiments have been disclosed, other aspects and embodiments will be apparent to those skilled in the art, and many changes and modifications can be made without departing from the spirit of the application, which is intended to be within the scope of the invention. The various aspects and embodiments disclosed herein are for illustration only and are not intended to limit the application, the actual scope of which is subject to the claims.
Claims (10)
1. A landscaping growth monitoring method based on high-resolution unmanned aerial vehicle remote sensing is characterized by comprising the following steps:
step1, acquiring basic image data, and acquiring remote sensing images in the urban landscaping range by using an unmanned aerial vehicle mounting camera.
Step2: and processing the basic image data, and carrying out orthographic correction, image stitching, noise processing and image enhancement on the obtained basic remote sensing image data.
Step3: afforestation object annotation data and field investigation data are obtained, the landscaping object comprises trees, shrubs and ground cover, the landscaping object annotation data comprise landscaping object categories and geographic position information, the field investigation data comprise tree height information, tree crown form information, blade color information, blade area index information, grassland coverage information and soil condition information, the landscaping objects in the processed image data are subjected to frame selection annotation, the categories and the geographic position information of the landscaping objects are recorded, and the field investigation is carried out on the landscaping objects based on the geographic position information to obtain the field investigation data.
Step4, obtaining landscaping object feature data based on landscaping object marking data and field investigation data, wherein the landscaping object feature data comprises morphological features, texture features, color features and spatial relationship features, the morphological features comprise tree crown shapes, trunk diameters and tree heights, the texture features comprise leaf textures, shrub textures and grassland textures, the color features comprise leaf colors, shrub colors and grassland colors, and the spatial relationship features comprise tree relative positions and distribution rules.
And Step5, integrating and forming a data set based on the characteristic data of the landscaping objects and the field investigation data, wherein the data set comprises a training set and a testing set, and constructing a landscaping growth analysis model based on the training set.
And Step6, evaluating the landscaping growth analysis model by using a test set, wherein evaluation data comprise calculation accuracy, recall rate indexes and prediction results of the visual model.
2. The method for monitoring landscaping growth based on remote sensing of high-resolution unmanned aerial vehicle according to claim 1, further comprising the Step of, in Step5, training the landscaping growth analysis model with the training set and evaluating the performance of the landscaping growth analysis model with the testing set.
3. The method for monitoring landscaping growth based on remote sensing of high-resolution unmanned aerial vehicle according to claim 2, wherein the training set accounts for 70% and the testing set accounts for 30%.
4. The method for monitoring the landscaping growth condition based on the high-resolution unmanned aerial vehicle remote sensing according to claim 1, wherein in Step3, the method further comprises the steps of adopting average statistics of other healthy trees of the same variety and the same age by adopting equal varieties, wherein the average statistics of other healthy trees of the same age are better and worse than the average, if the average is higher than the average, the blade color information is the color of the tree blade, the color of the tree blade is usually light green or bright green, the blade represents good growth condition, if the blade represents yellow or withered yellow, the blade may represent problems of poor growth or water shortage and the like, the blade Area Index information (LAI) is the ratio of the coverage Area of the blade to the ground Area, which is used for evaluating the degree of fullness and growth condition of the tree blade, the higher LAI value usually represents good growth condition, the general LAI value ranges from 1 to 10, the larger the value represents better growth condition, but the relation between the specific value and plant growth condition cannot be completely understood, the best discrimination standard is that compared with the average value of the LAI of healthy vegetation of the same variety, the coverage information of grassland is artificially good, the grassland coverage is used for evaluating the growth condition, a higher coverage generally indicates that the grass is growing well, a value of grass coverage between 0 and 1, 0 indicates worst, 1 indicates best, and soil condition information is soil conditions for judging the growth conditions, and in a better grass growth condition, the soil should be covered with effective vegetation without bare soil exposure.
5. The method for monitoring the growth of landscaping based on remote sensing of a high-resolution unmanned aerial vehicle according to claim 1, further comprising the Step of calculating the shape of the crown in the characteristic data of the landscaping object in such a way that the Crown Width (CW) is the maximum extension of the crown, and generally measuring the east-west width (EW) and the north-south width (NS) of the crown, wherein the calculation formula is: CW= (EW+NS)/2, the Crown Area (CA) refers to the coverage area of the crown vertical projection ground, the texture characteristic index and the calculation formula.
6. The method for monitoring landscaping growth based on remote sensing of high-resolution unmanned aerial vehicle as claimed in claim 5, wherein the calculation mode of the texture features in the feature data of the landscaping object is Entropy (Entropy): to describe the complexity and texture richness of the image.
The calculation formula is as follows: entropy= Sigma (P (i) ×log2 (P (i))), where P (i) is the probability of gray level per pel, contrast (Contrast): to describe the degree of gray scale difference between adjacent pixels in an image. The calculation formula is as follows: contrast= Σ ((i- μ) ≡2×P (i)), where i is the gray level of each pixel, μ is the average gray level of the image, energy (Energy): describing the texture details of the image, and calculating the following formula: energy= Σ (P (i)/(2), where P (i) is the probability of each pixel, correlation): to describe the correlation between the gray level of a picture element and neighboring picture elements in an image. The calculation formula is as follows: corelation= Σ ((i- μ) (j- μ) P (i, j)/(σ_i×σ_j), where i and j are gray levels of adjacent pixels, P (i, j) is a joint probability of adjacent pixels, μ is an average gray level of an image, σ is a standard deviation of gray levels, uniformity (Uniformity): the method is used for describing the consistency between pixel gray levels in the image, and the calculation formula is as follows: uniformity= Σ (P (i)/(2), where P (i) is the gray level probability of each pixel, degree of freedom (Degree of Freedom): the degree of gray level variation near the picture elements in the image is described. The calculation formula is as follows: degree of Freedom = Σ ((i-j)/(2 x P (i, j))/(P (i, j)), where i and j are the gray levels of neighboring pixels,
p (i, j) is the joint probability of neighboring pixels.
7. The method for monitoring the landscaping growth conditions based on the remote sensing of the high-resolution unmanned aerial vehicle according to claim 6, wherein the Color feature indexes and the calculation modes in the feature data of the landscaping object are Average colors (Average colors): color features are described by calculating average color values for pixels within an object region. An average value in a Color space such as RGB, HSV, lab and a Color Contrast (Color Contrast) can be calculated: to describe the degree of difference between the object region color and the background color in the image. The difference between the color histogram of the object region and the background color histogram, color uniformity (ColorUniformity), may be calculated: to describe uniformity of color and uniformity of distribution of image areas. The variance or standard deviation of the color histogram may be calculated to obtain the color uniformity index.
8. The method for monitoring landscaping growth based on remote sensing of high-resolution unmanned aerial vehicle according to claim 1, further comprising the Step of using a convolutional long-short-Term Memory recurrent neural network model (Convolutional Long Short-Term Memory, convLSTM) as a landscaping growth analysis model in Step 5.
9. The method for monitoring landscaping growth based on remote sensing of high-resolution unmanned aerial vehicle according to claim 8, wherein the landscaping growth analysis model comprises a network model input layer, a convolution layer, a hidden layer and an output layer, wherein the input layer is used for receiving basic remote sensing image data and field investigation data of each period and converting the basic remote sensing image data and the field investigation data into data types which can be identified by the landscaping growth analysis model, the convolution layer is used for extracting features from the basic remote sensing image data, the hidden layer is composed of a plurality of ConvLSTM units, and the hidden layer is used for learning and memorizing long-term dependency data by using forgetting gates, input gates and output gates and internal states in each ConvLSTM unit; the output layer is used for converting the output data of the hidden layer into a growth analysis result.
10. A system for monitoring landscaping growth based on high-resolution unmanned aerial vehicle remote sensing, which comprises a basic image data module, an image data processing module, a landscaping object labeling data module, an field investigation data module, a landscaping object characteristic data module, a data set module and a landscaping growth analysis model,
and a basic image data module: the remote sensing image acquiring method is used for acquiring a remote sensing image in the urban landscaping range by using the unmanned aerial vehicle mounting camera;
the image data processing module: the method is used for processing the basic image data, and carrying out orthographic correction, image stitching, noise processing and image enhancement on the acquired basic remote sensing image data;
afforestation object label data module: the method comprises the steps of performing frame selection labeling on landscaping objects in processed image data, and recording category and geographic position information of the landscaping objects;
field investigation data module: the method comprises the steps of performing field investigation on landscaping objects based on geographic position information to obtain field investigation data;
the landscaping object feature data module is used for obtaining landscaping object feature data based on output data of the landscaping object annotation data module and output data of the field investigation data module, the landscaping object feature data comprises morphological features, texture features, color features and spatial relationship features, the morphological features comprise crown shapes, trunk diameters and tree heights, the texture features comprise leaf textures, shrub textures and grassland textures, the color features comprise leaf colors, shrub colors and grassland colors, and the spatial relationship features comprise tree relative positions and distribution rules;
the data set module is used for integrating the landscaping object characteristic data module and the field investigation data module to output data, and comprises a training set module and a testing set module;
the training set module is used for constructing an landscaping growth analysis model, and the testing set module is used for evaluating the landscaping growth analysis model, wherein evaluation data comprise calculation accuracy, recall rate indexes and prediction results of the visualization model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311485354.5A CN117292282A (en) | 2023-11-09 | 2023-11-09 | Afforestation growth condition monitoring method and system based on high-resolution unmanned aerial vehicle remote sensing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311485354.5A CN117292282A (en) | 2023-11-09 | 2023-11-09 | Afforestation growth condition monitoring method and system based on high-resolution unmanned aerial vehicle remote sensing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117292282A true CN117292282A (en) | 2023-12-26 |
Family
ID=89251949
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311485354.5A Pending CN117292282A (en) | 2023-11-09 | 2023-11-09 | Afforestation growth condition monitoring method and system based on high-resolution unmanned aerial vehicle remote sensing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117292282A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115810151A (en) * | 2022-12-12 | 2023-03-17 | 江西省地质局地理信息工程大队 | Deep learning woodland extraction tree species recognition algorithm model |
| CN116258956A (en) * | 2022-12-12 | 2023-06-13 | 深圳市宇驰环境技术有限公司 | Unmanned aerial vehicle tree recognition method, unmanned aerial vehicle tree recognition equipment, storage medium and unmanned aerial vehicle tree recognition device |
| CN116797952A (en) * | 2023-07-07 | 2023-09-22 | 华南农业大学 | Litchi yield prediction method based on unmanned aerial vehicle multispectral image and growth information |
| CN116805396A (en) * | 2023-08-24 | 2023-09-26 | 杭州稻道农业科技有限公司 | Satellite remote sensing-based farmland weed precise identification method and device |
| WO2023197496A1 (en) * | 2022-04-15 | 2023-10-19 | 石河子大学 | Comprehensive evaluation indicator monitoring and evaluation method and system for machine-harvested cotton defoliation effects |
| CN116912578A (en) * | 2023-07-17 | 2023-10-20 | 北京师范大学 | Crop classification method, system and electronic equipment |
-
2023
- 2023-11-09 CN CN202311485354.5A patent/CN117292282A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023197496A1 (en) * | 2022-04-15 | 2023-10-19 | 石河子大学 | Comprehensive evaluation indicator monitoring and evaluation method and system for machine-harvested cotton defoliation effects |
| CN115810151A (en) * | 2022-12-12 | 2023-03-17 | 江西省地质局地理信息工程大队 | Deep learning woodland extraction tree species recognition algorithm model |
| CN116258956A (en) * | 2022-12-12 | 2023-06-13 | 深圳市宇驰环境技术有限公司 | Unmanned aerial vehicle tree recognition method, unmanned aerial vehicle tree recognition equipment, storage medium and unmanned aerial vehicle tree recognition device |
| CN116797952A (en) * | 2023-07-07 | 2023-09-22 | 华南农业大学 | Litchi yield prediction method based on unmanned aerial vehicle multispectral image and growth information |
| CN116912578A (en) * | 2023-07-17 | 2023-10-20 | 北京师范大学 | Crop classification method, system and electronic equipment |
| CN116805396A (en) * | 2023-08-24 | 2023-09-26 | 杭州稻道农业科技有限公司 | Satellite remote sensing-based farmland weed precise identification method and device |
Non-Patent Citations (1)
| Title |
|---|
| 田晓磊: "基于多源遥感数据的城市绿量测算——以银川市为例", 《中国优秀硕士学位论文全文数据库 基础科学辑 2023年第01期》, 15 January 2023 (2023-01-15) * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102855485B (en) | The automatic testing method of one grow wheat heading | |
| CN111368736A (en) | Rice refined estimation method based on SAR and optical remote sensing data | |
| CN110363246A (en) | A kind of fusion method of high-spatial and temporal resolution vegetation index NDVI | |
| CN111985724B (en) | Crop yield estimation method, device, equipment and storage medium | |
| CN108776106A (en) | A kind of crop condition monitoring method and system based on unmanned plane low-altitude remote sensing | |
| CN105513096A (en) | Method for estimating biomass of winter wheat | |
| CN102663396B (en) | Method for automatically detecting rice milky ripe stage | |
| CN112836725A (en) | Weak supervision LSTM recurrent neural network rice field identification method based on time sequence remote sensing data | |
| CN113963260A (en) | Extraction method and device for winter wheat planting area and computer equipment | |
| CN116227758B (en) | Agricultural product maturity prediction method and system based on remote sensing technology and deep learning | |
| CN110705182A (en) | Crop breeding adaptive time prediction method coupling crop model and machine learning | |
| CN112580671A (en) | Automatic detection method and system for multiple development stages of rice ears based on deep learning | |
| CN115456476A (en) | Territorial space planning data acquisition and analysis system based on machine vision | |
| CN112861712B (en) | Agricultural pest and disease monitoring method based on artificial intelligence and multi-temporal remote sensing | |
| CN113570273A (en) | Spatial method and system for irrigation farmland statistical data | |
| CN116482041B (en) | Rice heading period nondestructive rapid identification method and system based on reflection spectrum | |
| CN111191543A (en) | Rape yield estimation method | |
| CN117292282A (en) | Afforestation growth condition monitoring method and system based on high-resolution unmanned aerial vehicle remote sensing | |
| CN116308866A (en) | Rice ear biomass estimation method and system based on canopy reflection spectrum | |
| CN114140695B (en) | Prediction method and system for tea tree nitrogen diagnosis and quality index determination based on unmanned aerial vehicle multispectral remote sensing | |
| CN116399401A (en) | Agricultural planting system and method based on artificial intelligence | |
| CN114612794A (en) | Remote sensing identification method for land covering and planting structure in finely-divided agricultural area | |
| CN114299379A (en) | Shadow area vegetation coverage extraction method based on high dynamic image | |
| CN109241866B (en) | Automatic crop classification method and device based on historical crop distribution map | |
| CN113762768B (en) | Agricultural drought dynamic risk assessment method based on natural gas generator and crop model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |