An automatic identification method for Mikania micrantha in forest areas combining laser LiDAR data and aerial image data

By combining laser LiDAR data and aerial image data, features are extracted and the rule set of vegetation recognition is constructed, which solves the problems of low efficiency and insufficient accuracy of traditional vegetation and vegetation monitoring, and realizes efficient and accurate automatic identification and distribution recognition of vegetation and vegetation.

CN112651312BActive Publication Date: 2024-11-26BEIJING FORESTRY UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202011479538.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-12-15
Publication Date
2024-11-26
Estimated Expiration
2040-12-15

AI Technical Summary

Technical Problem

Traditional cymbidium monitoring and prediction methods are time-consuming and labor-intensive, inefficient, and the limited number of factors leads to a decrease in recognition accuracy, making it difficult to accurately identify and predict the explosion area of ​​cymbidium.

Method used

Combining laser LiDAR data and aerial image data, a multi-scale segmentation method for object-oriented, extracting spectral features, texture features, shape features and vegetation pigment ratio index, constructing a classification rule set of vacuma recognition, and using an object-oriented rule-based classification method for automatic Identify.

Benefits of technology

It effectively solves the problems of high labor intensity and low efficiency of manual investigation, improves the automatic identification accuracy and efficiency of vacuma, and can intuitively display the spatial distribution and explosion area of ​​vacuma.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN112651312B_ABST
    Figure CN112651312B_ABST
Patent Text Reader

Abstract

The present invention discloses a method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data, and belongs to the field of intelligent processing of forestry remote sensing images. The present invention provides a method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data, and the extraction steps are: determining the optimal segmentation scale of Mikania micrantha in the image and segmenting it; using the texture features, spectral features and shape features of Mikania micrantha, combined with elevation data and the vegetation pigment ratio index calculated based on aerial images, the outbreak area of ​​Mikania micrantha is automatically identified and extracted, and the distribution identification range of Mikania micrantha is established. The invention can effectively solve the problems of high labor intensity, difficult conditions and low efficiency in traditional Mikania discovery and manual investigation. The present invention can realize the accurate identification of Mikania micrantha in a large range, and can serve the comprehensive prevention, control and monitoring of Mikania micrantha.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent processing of forestry remote sensing images, and more particularly to a method for automatically identifying micrantha in forest areas by combining laser LiDAR data and aerial image data. Background Art

[0002] Due to the development of economic globalization, exchanges between regions of the world are becoming more frequent. The phenomenon of alien species invasion exists in almost all countries and regions in the world. Due to the lack of natural control mechanisms, these invasive species often threaten the local natural ecosystems, causing significant economic and ecological losses to the invaded areas. The problem of biological invasion has now become one of the three most difficult environmental problems in the 21st century. Among these invasive species, Mikania micrantha is listed as one of the most harmful tropical and subtropical weeds in the world due to its strong survival and reproduction ability and strong destructive power to the native vegetation in the invaded areas, and has attracted widespread attention from researchers.

[0003] In order to effectively control the spread and diffusion of Mikania micrantha, its existing spatial distribution must be identified and its dynamic changes must be monitored and predicted. The traditional monitoring and prediction method is to conduct manual field surveys and establish a diffusion model based on historical data. However, this method is not only laborious and time-consuming, but also cannot guarantee timeliness and accuracy. This requires the use of aerial remote sensing technology to systematically analyze the spatial distribution characteristics of Mikania micrantha and extract the outbreak area. At present, the MaxEnt model is mostly used for the identification of Mikania micrantha. However, since the growth of Mikania micrantha is affected by many factors, such as nutrients, water, light, slope, forest resource distribution, etc., the real-time data of these factors are difficult to obtain and fully grasp in time. Therefore, when the number of factors is limited, the accuracy of the model will be reduced. How to accurately identify and predict the outbreak area of ​​Mikania micrantha based on existing data is a key issue that needs to be solved at present. Based on this, a large-scale automatic identification method of Mikania micrantha is studied, which is of great significance for the prevention and control of Mikania micrantha and the protection of forest resources. Summary of the invention

[0004] The purpose of the present invention is to provide a method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data, based on geographic information system (GIS), remote sensing (RS), and computer (CS) technology, combining airborne laser radar data and aerial image data and applying them to spatial information analysis technology for forest pest control, which is used to judge the scope and degree of forest damage caused by Mikania micrantha and other similar harmful organisms, and can assist forestry departments in carrying out corresponding prevention and control work. The invention can effectively solve the problems of high labor intensity, difficult conditions, and low efficiency in traditional Mikania discovery and manual investigation; it can also effectively overcome the shortcomings of low resolution of ordinary optical images and low accuracy of identifying Mikania micrantha only by optical band combination.

[0005] In order to achieve the above object, the present invention adopts the following technical solution:

[0006] A method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data comprises the following steps:

[0007] S1. Use the object-oriented multi-scale segmentation method to segment the high-resolution aerial images of the study area, determine the segmentation scale and segmentation scale parameters of Mikania micrantha, and obtain the object-based multispectral image of the study area;

[0008] S2, extracting the spectral features, texture features, shape features, elevation and vegetation pigment ratio index of the multispectral image as characteristic indicators, screening them through variance analysis and normal distribution test, and obtaining the characteristic combination for automatic identification of Mikania micrantha outbreak areas;

[0009] S3. Construct a Mikania micrantha identification and classification rule set based on the selected feature combination, and use an object-oriented rule-based classification method to extract the Mikania micrantha outbreak invasion area from the image object block to obtain the Mikania micrantha distribution identification range.

[0010] Preferably, in step S1, the segmentation scale is calculated by using the segmentation scale traversal ESP algorithm;

[0011] The segmentation scale is 60, 76, 80, 100, 114, 135, 148, 168, 188, 201, 208, 217 or 245.

[0012] Preferably, in step S1, the segmentation scale parameters include a shape parameter, a compactness parameter and a segmentation scale;

[0013] Among them, the shape parameter is 0.5, the compactness parameter is 0.5, and the segmentation scale is 100.

[0014] Preferably, in step S2, a digital elevation model (DEM) is extracted by using the airborne laser point cloud data through a Kriging interpolation (Kriging) algorithm.

[0015] Preferably, in step S2, the spectral characteristics include the average value, brightness and vegetation index of pixels in red, green and blue bands;

[0016] Preferably, the average value is calculated by formula (1):

[0017]

[0018] Where: is the average value of the pixels constituting an image object on the upper layer, L = 1, 2, ..., m; n is the number of pixels; is the value of the i-th pixel on the top layer.

[0019] Preferably, the brightness is calculated by formula (2):

[0020]

[0021] In the formula, is the brightness; m is the number of layers of the image object; is the average value of the Lth layer value containing spectral information.

[0022] Preferably, the vegetation index is calculated by formula (3):

[0023] VI=(2G'-R'-B')-(1.4R'-G') (3)

[0024]

[0025] Where VI is the vegetation index, and R, G, and B are the average values ​​of the red, green, and blue bands, respectively.

[0026] Preferably, in step S2, the texture features include the dissimilarity and correlation of the gray level co-occurrence matrix;

[0027] The dissimilarity is calculated by formula (4):

[0028]

[0029] In the formula, G dis is the difference; i, j are the number of rows and columns respectively; P i,j is the normalized value of the i-th and j-th elements, and N is the number of rows or columns;

[0030] The correlation is calculated by formula (5):

[0031] G cor =[∑ i ∑ j (iu x )(ju y )] / (σ x σ y ) (5)

[0032] In the formula, G cor is the difference; i, j are the number of rows and columns respectively; u x 、u y are the means of rows and columns respectively; σ x , σ y are the standard deviations of the rows and columns, respectively.

[0033] Preferably, in step S2, the shape feature includes an aspect ratio;

[0034] The aspect ratio is calculated by formula (6):

[0035]

[0036] Where Y is the aspect ratio, w is the width, l is the length, s is the covariance matrix, eig1(s), eig2(s) are the covariance eigenvalues;

[0037] The vegetation pigment ratio index is calculated by formula (7):

[0038] PPR = (gb) / (g+b) (7)

[0039] Wherein, PPR is the vegetation pigment ratio index; g is the green band DN value; b is the blue band DN value.

[0040] Preferably, it is characterized in that, in step S3, a threshold classification algorithm is used to adjust the threshold range of the feature combination, identify and mark Mikania micrantha, and determine a Mikania micrantha identification and classification rule set;

[0041] The Mikania micrantha identification and classification rule set is: scale=100, shape=0.5, compactness=0.5, 29<DEM<329, mean layer2>175, 0.15<PPR<0.25.

[0042] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data, which has the following technical effects:

[0043] (1) The present invention provides a method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data, which can effectively solve the problems of high labor intensity, difficult working conditions, and low efficiency when conducting manual surveys of Mikania micrantha in large-scale forest farms. It can realize automatic identification of Mikania micrantha based on high-resolution images, and greatly improve the detection of invasive organisms.

[0044] (2) The present invention provides a method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data, and constructs a set of automatic identification rules for Mikania micrantha outbreak areas, which can effectively extract Mikania micrantha outbreak areas and intuitively display the spatial range of Mikania micrantha's climbing growth.

[0045] (3) The present invention provides a method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data, which can effectively overcome the problems of complex and slow processing in traditional data processing and achieve rapid identification of Mikania micrantha in a large range and with high precision. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0047] Figure 1 The accompanying drawing is a technical roadmap of a method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0049] The embodiment of the present invention provides a method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data. Figure 1 As shown, the following steps are included:

[0050] S1: Image segmentation of the original multispectral image, using the object-oriented multi-scale segmentation method to segment the high-resolution aerial image of the study area, determine the optimal segmentation scale and parameters of Mikania micrantha, and obtain the object-based multispectral image of the study area;

[0051] S2: Extract spectral features (average value, brightness, vegetation index, vegetation pigment ratio index), texture features (dissimilarity and correlation of gray-level co-occurrence matrix), shape features (aspect ratio) and elevation data of multispectral images as feature indicators, extract DEM using airborne laser point cloud data, and screen through variance analysis and normal distribution test to obtain feature combinations for automatic identification of Mikania outbreak areas;

[0052] S3: Construct a Mikania micrantha identification rule set based on the selected feature combination, and use an object-oriented rule-based classification method to extract the Mikania micrantha outbreak invasion area from the image object block to obtain the Mikania micrantha distribution identification range.

[0053] Example 1

[0054] This embodiment provides a method for identifying Mikania micrantha by combining laser LiDAR data and aerial images, comprising the following steps:

[0055] S1: Determine the optimal segmentation scale of Mikania micrantha and obtain the image object of the study area;

[0056] (1) This study used the segmentation scale traversal ESP (Estimation of Scale Parameter) algorithm to calculate the optimal segmentation scale. The initial segmentation scale was set to 1, the step size was 10, and the number of cycles was 200. The segmentation scales at which the peaks appeared were 60, 76, 80, 100, 114, 135, 148, 168, 188, 201, 208, 217, or 245. After segmentation comparison one by one, it was determined that the shape parameter 0.5, the compactness parameter 0.5, and the segmentation scale parameter 100 were the optimal segmentation scale parameters for identifying Mikania objects.

[0057] (2) Based on the above optimal segmentation scale parameters, a multi-scale segmentation algorithm is used to obtain the image object of the study area;

[0058] S2: Extract the feature combination for automatic identification of Mikania micrantha outbreak areas;

[0059] (1) After object-oriented multi-scale segmentation of high-resolution helicopter aerial images of the study area, the spectral characteristics, texture characteristics, shape characteristics, elevation and vegetation pigment ratio index (PPR) of the object were extracted based on the following formula:

[0060] 1) Spectral characteristics include the average value of pixels in the three bands of red, green and blue (Formula 1), brightness (Formula 2), and VI value (Formula 3);

[0061]

[0062] Where: is the average value of the pixels constituting an image object on the upper layer, L = 1, 2, ..., m; n is the number of pixels; is the value of the i-th pixel on the top layer, also called the layer value.

[0063]

[0064] Where: is the brightness; m is the number of layers of the image object; is the average value of the Lth layer value containing spectral information.

[0065] VI=(2G'-R'-B')-(1.4R'-G') (3)

[0066]

[0067] Where: VI is the vegetation index, R, G, B are the average values ​​of the red, green and blue bands respectively.

[0068] 2) Texture features include the commonly used gray level co-occurrence matrix dissimilarity (Formula 4) and correlation (Formula 5);

[0069]

[0070] Where: G dis is the difference; i, j are the number of rows and columns respectively; is the standardized value of the i, jth element, and N is the number of rows or columns.

[0071] G cor =[∑ i ∑ j (iu x )(ju y )] / (σ x σ y ) (5)

[0072] Where: G cor is the correlation; i, j are the number of rows and columns respectively; u x 、u y are the means of rows and columns respectively; σ x , σ y are the standard deviations of the rows and columns, respectively.

[0073] 3) Shape feature selection aspect ratio (Formula 6);

[0074]

[0075] Where: Y is the aspect ratio, w is the width, l is the length, s is the covariance matrix, eig1(s), eig2(s) are the covariance eigenvalues.

[0076] 4) Vegetation pigment ratio index (Formula 7);

[0077] PPR = (gb) / (g+b) (7)

[0078] Wherein, PPR is the vegetation pigment ratio index; g is the green band DN value; b is the blue band DN value.

[0079] 5) Use airborne laser point cloud data to extract the digital elevation model (DEM) through the Kriging interpolation algorithm.

[0080] (2) The extracted characteristic indicators were subjected to variance analysis and kurtosis coefficient normal distribution test, and the indicators with small variance and large correlation were removed. The G band average, brightness, DEM and vegetation pigment ratio index (PPR) indicators with large variance and rich information were retained to establish a feature combination for automatic identification of Mikania outbreak areas;

[0081] Step 3: Establish the automatic identification range of Mikania micrantha outbreak areas;

[0082] (1) Based on the feature combination selected in Step 2, the information contained in each feature is used to create an information knowledge base of Mikania micrantha objects. The threshold classification (AssignClass) algorithm is used to continuously adjust the threshold range of the feature combination to identify and label Mikania micrantha. The classification rule set of Mikania micrantha is determined as follows: scale = 100, shape = 0.5, compactness = 0.5, 29 < DEM < 329, 190 < brightness ≤ 220, meanlayer2 > 175, 0.15 < PPR < 0.25;

[0083] According to the created rule set, an object-oriented rule-based classification method was used to separate Mikania micrantha from climbing plants, and the automatic identification range of Mikania micrantha outbreak area was obtained.

[0084] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0085] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data. It is characterized in that The following steps are involved: S1. Use the object-oriented multi-scale segmentation method to segment the high-resolution aerial images of the study area, determine the segmentation scale and segmentation scale parameters of Mikania micrantha, and obtain the object-based multispectral image of the study area; S2, extracting the spectral features, texture features, shape features, elevation and vegetation pigment ratio index of the multispectral image as characteristic indicators, screening them through variance analysis and normal distribution test, and obtaining the characteristic combination for automatic identification of Mikania micrantha outbreak areas; S3, constructing a Mikania micrantha identification and classification rule set based on the selected feature combination, and extracting the Mikania micrantha outbreak invasion area from the image object block by using an object-oriented rule-based classification method to obtain the Mikania micrantha distribution identification range; In step S2, the texture features include the dissimilarity and correlation of the gray level co-occurrence matrix; The dissimilarity is calculated by formula (4): In the formula, G dis for heterogeneity; i, j are the number of rows and columns respectively; P i,j is the normalized value of the i-th and j-th elements, and N is the number of rows or columns; The correlation is calculated by formula (5): G cor =[∑ i ∑ j (iu x (ju) y )] / (σ x s y ) (5) In the formula, G cor is the difference; i, j are the number of rows and columns respectively; u x 、u y are the means of rows and columns respectively; σ x , σ y are the standard deviations of rows and columns, respectively; The shape characteristics include aspect ratio; The aspect ratio is calculated by formula (6): Where Y is the aspect ratio, w is the width, l is the length, s is the covariance matrix, eig 1 (s), eig 2 (s) is the covariance eigenvalue; The vegetation pigment ratio index is calculated by formula (7): PPR = (gb) / (g+b) (7) In the formula, PPR is the vegetation pigment ratio index; g is the DN value of the green band; b is the DN value of the blue band; The elevation is obtained by extracting a digital elevation model using a Kriging interpolation algorithm using airborne laser point cloud data; In step S2, variance analysis and kurtosis coefficient normal distribution test are performed on the extracted characteristic indicators to establish a characteristic combination for automatic identification of Mikania micrantha outbreak areas; In step S3, according to the selected feature combination, the information contained in each feature is used to create an information knowledge base of Mikania micrantha objects, and a threshold classification algorithm is used to continuously adjust the threshold range of the feature combination to identify and label Mikania micrantha, and determine a Mikania micrantha classification rule set; According to the created rule set, an object-oriented rule-based classification method was used to separate Mikania micrantha from climbing plants, and the automatic identification range of Mikania micrantha outbreak area was obtained.

2. The method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data according to claim 1, It is characterized in that In step S1, the segmentation scale algorithm is the ESP algorithm; The segmentation scale is one of 60, 76, 80, 100, 114, 135, 148, 168, 188, 201, 208, 217 and 245.

3. The method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data according to claim 1, It is characterized in that In step S1, the segmentation scale parameters include shape parameters, compactness parameters and segmentation scale; Among them, the shape parameter is 0.5, the compactness parameter is 0.5, and the segmentation scale is 100.

4. The method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data according to claim 1, It is characterized in that In step S2, the spectral characteristics include the average value, brightness and vegetation index of pixels in the red, green and blue bands.

5. The method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data according to claim 4, It is characterized in that The average value is calculated by formula (1): Where: is the average value of the pixels constituting an image object on the Lth layer, L = 1, 2, ..., m; n is the number of pixels; is the value of the i-th pixel on the L-th layer.

6. The method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data according to claim 4, It is characterized in that The brightness is calculated by formula (2): In the formula, is the brightness; m is the number of layers of the image object; It is the average value of the pixels constituting an image object on the Lth layer.

7. The method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data according to claim 4, It is characterized in that The vegetation index is calculated by formula (3): VI=(2G'-R'-B')-(1.4R'-G') (3) Where VI is the vegetation index, and R, G, and B are the average values ​​of the red, green, and blue bands, respectively.

8. The method for automatically identifying Mikania micrantha in forest areas by combining laser LiDAR data and aerial image data according to claim 1, It is characterized in that In step S3, a threshold classification algorithm is used to adjust the threshold range of the feature combination, identify and mark Mikania micrantha, and determine a Mikania micrantha identification and classification rule set; The Mikania micrantha identification and classification rule set is: scale = 100, shape = 0.5, compactness = 0.5, 29 <DEM<329, mean layer2>175, 0.15 <PPR<0.25。