KR102625621B1 - Customized vegetation index measurement method based on UAV multispectral sensor considering the characteristics of land cover in urban areas - Google Patents

Abstract

A customized vegetation index measurement method based on UAV multispectral sensors considering land cover characteristics of urban areas is disclosed. Satellite images have been widely used in the fields of urban planning, landscape management, and environmental monitoring, but rapid filming is difficult and expensive. Recently, imaging technology using unmanned aerial vehicles (UAVs) equipped with multispectral cameras has been introduced. . Previously, vegetation information over a wide area was acquired using multi-spectral bands of satellites, but recently, UAV technology that can obtain vegetation information in a faster and more accurate manner is being actively researched. This study acquired multispectral images using a UAV equipped with a Micasense RedEde MX camera and analyzed the Normalized Difference Vegetation Index (NDVI), Green Normalized Difference Vegetation Index (GNDVI), Blue Normalized Difference Vegetation Index (BNDVI), and RGBVI ( Six existing vegetation indices such as Red Green Blue Vegetation Index (GRVI), Green Red Vegetation Index (GRVI), and Soil Adjusted Vegetation Index (SAVI) were analyzed. However, in the process of analyzing vegetation in urban areas using existing vegetation indices, there were cases where the vegetation index values of steel plate roofs, waterproof coated roofs, and urethane coated areas were similar to or slightly higher than those of vegetation areas such as grass. To improve the vegetation misclassification problem, various types of vegetation index equations were developed using multispectral images captured by UAV. As a result of performing Kappa coefficient analysis on the developed vegetation index, the squared Red-Blue NDVI index showed the best results when analyzing vegetation that reflects the land cover of urban areas. The new vegetation index (calculation 3) developed in this study will be very useful in effectively classifying vegetation in urban areas with various types of land cover, such as steel plate roofs, waterproof coating roofs, and basketball courts constructed with urethane coating.

Description

Customized vegetation index measurement method based on UAV multispectral sensor considering the characteristics of land cover in urban areas}

The present invention relates to a customized vegetation index measurement method based on a UAV multi-spectral sensor considering land cover characteristics of urban areas. More specifically, to accurately analyze vegetation distribution areas reflecting urban area cover characteristics, steel plate roofs, waterproof coating roofs, etc. In addition, new vegetation index calculation formulas (1), (2), and (3) were developed to solve the problem of misclassification of vegetation in urban areas in the urethane coating area and classify accurate vegetation areas by considering the cover characteristics of urban areas. This study is about a customized vegetation index measurement method based on a UAV multi-spectral sensor that analyzes the vegetation of urban areas with cover and considers land cover characteristics in urban areas.

1. Background

The frequency and intensity of heat waves are increasing worldwide, resulting in negative impacts on human health (Fahad et al., 2019; Sabrin et al., 2020; Uddin et al., 2019). The heat island effect in urban areas that occurs locally reduces the cooling effect due to impervious pavement and deforestation, which releases artificial heat and causes high concentrations of air pollutants (Peng et al., 2020). Recently, as urbanization accelerates, impermeable surfaces such as buildings, roads, and pavements dominate urban spaces, increasing surface temperatures compared to surrounding rural areas due to their ability to absorb and retain heat (Voogt·Oke, 2003 ; Sabrin et al., 2021; Ziter et al., 2019).

Vegetation generally plays a large role in regulating temperature (Ellison et al., 2017; Su et al., 2019). However, 88% of vegetation in urban areas has been destroyed and replaced by artificial surfaces during urbanization over the past few decades (d'Amour et al., 2016). Therefore, in order to create a comfortable residential environment, policies are needed to accurately identify the distribution characteristics of vegetation distributed in urban areas and to effectively manage and maintain it.

Vegetation monitoring research using remote sensing technology has been actively conducted in various fields such as urban planning, landscaping, forestry, water resources, and the environment (Jose·Paulo, 2010; Zhang et al., 2014). In particular, Vegetation Index (VI), which analyzes the spectroscopic characteristics of chlorophyll in plants based on optical and near-infrared bands, has been used as a representative indicator to evaluate vegetation distribution and growth characteristics in the field of remote sensing ( Na Sang-il et al., 2016; Tomas·Gausman, 1977). In particular, the Normalized Difference Vegetation Index (NDVI) published by Rouse et al. (1973) is most widely used in various fields.

Vegetation index research using remote sensing technology mainly uses MSS, TM, and ETM+ sensors mounted on Landsat satellites, and MODIS and SPOT images have also recently been used (Yeseul Kim et al., 2014; Jongmin Yeom et al., 2008; Myeonghee Jeong and Eunmi Jang) , 2013; Mireia et al., 2016; Ranjay et al., 2017; Shrestha et al., 2017; Zhou et al., 2020).

In addition to NDVI, vegetation indices are very diverse, including GNDVI (Green Normalized Difference Vegetation Index), BNDVI (Blue Normalized Difference Vegetation Index), RGBVI (Red Green Blue Vegetation Index), GRVI (Green Red Vegetation Index), and SAVI (Soil Adjusted Vegetation Index). (Genevieve et al., 1996; Xingwang·Yuanbo, 2016).

Remote sensing data from a specific period of time are needed to extract biophysical information of plants. Existing remote sensing research has mainly relied on satellite images, but due to the influence of periodic resolution, vegetation monitoring research based on unmanned aerial vehicles (UAVs) capable of capturing real-time images is currently actively underway (Sotille et al., 2020; Estrany et al. , 2019; Zhao et al., 2020).

Lee et al. (2016) estimated crop growth factors by extracting the NDVI vegetation index by shooting images acquired using a drone equipped with a multispectral camera, and Geunsang Lee and Yeonwoong Choi (2019) calculated NDVI, GNDVI, and NDVI calculated from drone images. A comparative study was conducted between the NDRE and SAVI vegetation indices and the vegetation indices calculated from locally observed spectral values. In addition, Francisco et al. (2015) conducted a study to monitor the growth status of sunflowers using time-series drone images, and Juan et al. (2013) analyzed the leaf area index of onions using drones. And, Juliane et al. (2015) conducted a study to estimate crop height from drone-based topographic data and vegetation index to monitor biomass.

In order to accurately classify vegetation areas during the image analysis process, the optimal threshold value must be selected from the calculated vegetation index section. The higher the vegetation index boundary value, the higher the chlorophyll level and canopy density of trees and grasslands, indicating an area with better vegetation vitality. Generally, in the field of urban planning and landscaping, areas with high vegetation vitality are designated as vegetation areas and used as data for complex design and construction. In addition, in the urban regeneration project, which has recently become an issue, urban forests and parks are being designed to build a pleasant urban environment such as reducing heat waves, and in the stage of verifying the effects of vegetation creation, the vegetation development project area is being designed to have a high level of vegetation vitality. Modeling is performed based on this consideration.

However, in the analysis of non-point pollution sources and soil loss, the role of plant leaves and branches is important, but the role of plant roots in preventing soil particles from escaping during actual rainfall is very important, so even grass and lawns with somewhat low vegetation vitality are also important. It is desirable to classify and model by vegetation area (Ganasri·Ramesh, 2016; Irina·Goga, 2018). Therefore, when performing modeling using a land cover map including vegetation, it is important to classify the vegetation area by determining the vegetation vitality that matches the business characteristics.

As prior art related to this, patent registration number 10-17281370000 discloses "Image classification method by land cover item using satellite images and GIS", which is an image classification method by land cover item using satellite image (Landsat image) and GIS. To reflect the latest highly accurate land cover characteristics, a land cover map was constructed through image classification techniques for land cover classes of water bodies, forests, urban areas, agricultural lands (rice fields/fields), and artificial features for specific regions.

Recently, interest in urban regeneration projects is very high in Korea. Urban regeneration is a city that is declining due to population decline, changes in industrial structure, indiscriminate expansion of the city, and deterioration of the residential environment. It is economically, socially, and physically improved by strengthening the capabilities of the city according to local characteristics, introducing new functions, or utilizing local resources. ·This refers to environmentally revitalizing.

Among these, vegetation plays a role in providing a comfortable living environment and a space for rest from heat waves and the urban heat island phenomenon caused by climate change.

As a result of using the existing vegetation index formula in the process of producing vegetation drawings to be used in urban regeneration projects, it was confirmed that vegetation index values were relatively high in the areas of steel plate roofs, waterproof coating roofs, and urethane coating areas. If the range of vegetation is simply limited to trees with very high vegetation vitality, only areas with a high vegetation index boundary value are classified as vegetation, so steel plate roofs, waterproof coated roofs, and urethane coated areas are classified as vegetation areas. excluded. However, if grass or grassland with vegetation vitality above average is classified as a vegetation area, areas with a vegetation index boundary value above average must also be classified as vegetation, so steel plate roofs, waterproof coating roofs, and urethane coating areas are also classified as vegetation areas. A problem arises.

In urban areas, when using the existing six vegetation indices such as NDVI, NDVI, GNDVI, BNDVI, RGBVI, GRVI, and SAVI, a problem occurs in which steel sheet roofs, urethane coatings, and waterproof coating areas are misclassified as vegetation.

Patent registration number 10-17281370000 (registration date April 12, 2017), “Image classification method by land cover item using satellite images and GIS”, Patent holder: Halla Geographic Information Co., Ltd., Inventors: Geunsang Lee, Jeong Lee

d'Amour, C.B., Reitsma, F., Baiocchi, G., Barthel, S., Gneralp, B., Erb, K.-H., Haberl, H., Creutzig, F., 2017, Future urban land expansion and implications for global croplands, 114, pp.8939-8944. Ellison, D., Morris, C.E., Locatelli, B., Sheil, D., Cohen, J., Murdiyarso, D., Gutierrez, V., van Noordwijk, M., Creed, I.F., Pokorny, J., etalon017, Trees, forests and water: Cool insights for a hot world, Global Environmental Change-Human and Policy Dimensions 43, pp.51-61. Estrany, J., Ruiz, M., Calsamiglia, A., Carriqui, M., Garcia-Comendador, J., Nadal, M., Fortesa, J., Lopez-Tarazon, J.A., Medrano, H., Gago, J., Sediment connectivity linked to vegetation using UAVs: High-resolution imagery for ecosystem management, Science of the Total Environment 671, pp.1192-1205. Fahad, M.G.R., Nazari, R., Bhavsar, P., Jalayer, M., Karimi, M., 2019, A decision-support framework for emergency evacuation planning during extreme storm events, Transportation Research Part D-Transport and Environment 77, pp.589-605. Francisco Aguera Vega, Fernando Carvajal Ramirez, Monica Perez Saiz, and Francisco Orgaz Rosua, 2015, Multi-temporal imaging using an unmanned aerial vehicle for monitoring a sunflower crop, Biosystems Engineering I32, pp.19-27. Ganasri, B.P. and Ramesh, H., 2016, Assessment of soil erosion by RUSLE model using remote sensing and GIS, Geoscience Frontiers 7, pp.953-961. Genevieve Rondeaux, Mochael Steven, and Frederic Baret, 1996, Optimization of Soil-Adjusted Vegetation Indices, Remote Sensing of Environment 55, pp.95-107. Irina Khubulava and Goga Chakhaia, 2018, Simulative modeling of the soil erosion processes, Annuals of Agrarian Science 16(2), pp.185-188. Jose Pablo Solans Vila and Paulo Barbosa, 2010, Post-fire vegetation regrowth detection in the Deiva Marina region using Landsat TM and ETM+ data, Ecological Modeling 221, pp.75-84. Juan I. Corcoles, Jose F. Ortega, David Hernandez, and Miguel A. Moreno, 2013, Estimation of lear area index in onion using an unmanned aerial vehicle, Biosystems Engineering II5, pp.31-42. Juliane Bendig, Kang Yu, Helge Aasen, Andreas Bolten, Simon Bennertz, Janis Broscheit, Martin L. Gnyp, and Georg Bareth, 2015, Combinating UAV-based plant height from crop surface models, visible and near infrared vegetation indices for biomass monitoring in barley, International Journal of Applied Earth Observation and Geoinformation 39, pp.79-87. Lee, K.D., Lee, Y.E., Park, C.W., Hong, S.Y., and Na, S.I., 2016, Study on Reflectance and NDVI of Aerial Images using a Fixed-Wing UAV eBee, Korean Journal of Soil Science and Fertilizer 49(6), pp.731-742. Mireia G, Marta C, David M, Payam D, Antonia V, Antoni P, and Mark JN., 2016, Normalized difference vegetation index (NDVI) as a marker of surrounding greenness in epidemiological studies: The case of Barcelona city, Urban Forestry & Urban Greening 19, pp.88-94. Peng, J., Hu, Y., Dong, J., Liu, Q., Liu, Y., 2020, Quantifying spatial morphology and connectivity of urban heat islands in a megacity: A radius approach, Sci Total Environ 714, doi:10.1016/j.scitotenv.2020.136792. Ranjay S, Liping D, Eugene G.Y, Lingjun K, Yuan-zheng S, and Yu-qi B., 2017, Regression model to estimate flood impact on corn yield using MODIS NDVI and USDA cropland data layer, Journal of Integrative Agriculture 16, pp.398-407. Rouse, J., Haas, R.H., Schell, J.A., 1974, Monitoring vegetation systems in the Great Plains with ERTS 351, p.309. Sabrin, S., Karimi, M., Nazari, R., 2020, Developing Vulnerability Index to Quantify Urban Heat Islands Effects Coupled with Air Pollution: A Case Study of Camden, International Journal of Geo-Information 9, doi:ARTN 34910.3390/ijgi9060349. Shrestha, R., Di, L.P., Yu, E.G., Kang, L.J., Shao, Y.Z., Bai, Y.Q., 2017, Regression model to estimate flood impact on corn yield using MODIS NDVI and USDA cropland data layer, J Integr Agr 16, pp.398-407. Sotille, M.E., Bremer, U.F., Vieira, G., Velho, L.F., Petsch, C., Simoes, J.C., 2020, Evaluation of UAV and satellite-derived NDVI to map maritime Antarctic vegetation, Applied Geography 125, doi:ARTN 10232210.1016/j.apgeog.102322. Su, Y.X., Liu, L.Y., Wu, J.P., Chen, X.Z., Shang, J.L., Ciais, P., Zhou, G.Y., Lafortezza, R., Wang, Y.P., Yuan, W.P., etalon019, Quantifying the biophysical effects of forests on local air temperature using a novel three-layered land surface energy balance model, Environment International 132, doi:ARTN 10508010.1016/j.envint.2019.105080. Tomas, J.R. and H.W. Gausman, 1977, Leat reflectance vs. leaf chlorophyll and carotenoid concentrations for eight crops, Agronomy Journal 69, pp.799-802. Uddin, M.N., Islam, A.K.M.S., Bala, S.K., Islam, G.M.T., Adhikary, S., Saha, D., Haque, S., Fahad, M.G.R., Akter, R., 2019, Mapping of climate vulnerability of the coastal region of Bangladesh using principal component analysis, Applied Geography 102, pp.47-57. Voogt, J.A., Oke, T.R., 2003, Thermal remote sensing of urban climates, Remote Sensing of Environment 86, pp.370-384. Xingwang F. and Yuanbo L., 2016, A global study of NDVI difference among moderate-resolution satellite sensors. Journal of Photogrammetry and Remote Sensing 121, pp.177-191. Zhang Feng, Zhang Li-Wen, Shi Jing-Jing, and Huang Jing-Feng, 2014, Soil Moisture Monitoring Based on Land Surface Temperature Vegetation Index Space derived from MODIS Data. PEDOSPHERE 24(4):450-460. Zhao, F., Wu, X., Wang, S.J., Object-oriented Vegetation Classification Method based on UAV and Satellite Image Fusion 174, pp.609-615. Zhou, X.J., Wang, P.X., Tansey, K., Zhang, S.Y., Li, H.M., Wang, L., 2020, Developing a fused vegetation temperature condition index for drought monitoring at field scales using Sentinel-2 and MODIS imagery, Comput Electron Agr 168, doi:ARTN 10514410.1016/j.compag.2019.105144. Ziter, C.D., Pedersen, E.J., Kucharik, C.J., Turner, M.G., 2019, Scale-dependent interactions between tree canopy cover and impervious surfaces reduce daytime urban heat during summer, Proceedings of the National Academy of Sciences of the United States of America ,116, pp.7575-7580. Ye-seul Kim, No-wook Park, Seok-young Hong, Gyeong-do Lee, Hee-young Yoo, 2014, Early production of large-scale crop classification maps using time series vegetation index and past crop cultivation patterns, Journal of the Korean Society of Remote Sensing 30(4), pp.493-503. Sang-il Na, Chan-won Park, Young-geun Jeong, Cheon-sik Kang, In-bae Choi, Gyeong-do Lee, 2016, Selection of optimal vegetation index for remote sensing-based pulse crop estimation, Journal of the Korean Society of Remote Sensing 32(5), pp.483-497. Jong-min Yeom, Gyeong-su Han, Chang-seok Lee, Yun-young Park, Young-seop Kim, 2008, Detection of vegetation changes in North Korea using SPOT/VEGETATION NDVI data, Journal of the Korean Society of Geographic Information 11(2), pp.28-37. Geunsang Lee, Yeonwoong Choi, 2019, Analysis of agricultural land spectral characteristics and vegetation index using UAV, Journal of the Korean Society of Geographic Information 22(4), pp.86-101. Myunghee Jeong, Eunmi Jang, 2013, Detection of surface vegetation changes through harmonic analysis of MODIS NDVI time series data, Journal of the Korean Society of Remote Sensing 29(4), pp.351-360.

d'Amour, C.B., Reitsma, F., Baiocchi, G., Barthel, S., Gneralp, B., Erb, K.-H., Haberl, H., Creutzig, F., 2017, Future urban land expansion. and implications for global croplands, 114, pp.8939-8944. Ellison, D., Morris, C.E., Locatelli, B., Sheil, D., Cohen, J., Murdiyarso, D., Gutierrez, V., van Noordwijk, M., Creed, I.F., Pokorny, J., etalon017 , Trees, forests and water: Cool insights for a hot world, Global Environmental Change-Human and Policy Dimensions 43, pp.51-61. Estrany, J., Ruiz, M., Calsamiglia, A., Carriqui, M., Garcia-Comendador, J., Nadal, M., Fortesa, J., Lopez-Tarazon, J.A., Medrano, H., Gago; J., Sediment connectivity linked to vegetation using UAVs: High-resolution imagery for ecosystem management, Science of the Total Environment 671, pp.1192-1205. Fahad, MGR, Nazari, R., Bhavsar, P., Jalayer, M., Karimi, M., 2019, A decision-support framework for emergency evacuation planning during extreme storm events, Transportation Research Part D-Transport and Environment 77, pp.589-605. Francisco Aguera Vega, Fernando Carvajal Ramirez, Monica Perez Saiz, and Francisco Orgaz Rosua, 2015, Multi-temporal imaging using an unmanned aerial vehicle for monitoring a sunflower crop, Biosystems Engineering I32, pp.19-27. Ganasri, BP and Ramesh, H., 2016, Assessment of soil erosion by RUSLE model using remote sensing and GIS, Geoscience Frontiers 7, pp.953-961. Genevieve Rondeaux, Mochael Steven, and Frederic Baret, 1996, Optimization of Soil-Adjusted Vegetation Indices, Remote Sensing of Environment 55, pp.95-107. Irina Khubulava and Goga Chakhaia, 2018, Simulative modeling of the soil erosion processes, Annuals of Agrarian Science 16(2), pp.185-188. Jose Pablo Solans Vila and Paulo Barbosa, 2010, Post-fire vegetation regrowth detection in the Deiva Marina region using Landsat TM and ETM+ data, Ecological Modeling 221, pp.75-84. Juan I. Corcoles, Jose F. Ortega, David Hernandez, and Miguel A. Moreno, 2013, Estimation of rare area index in onion using an unmanned aerial vehicle, Biosystems Engineering II5, pp.31-42. Juliane Bendig, Kang Yu, Helge Aasen, Andreas Bolten, Simon Bennertz, Janis Broscheit, Martin L. Gnyp, and Georg Bareth, 2015, Combining UAV-based plant height from crop surface models, visible and near infrared vegetation indices for biomass monitoring in barley, International Journal of Applied Earth Observation and Geoinformation 39, pp.79-87. Lee, KD, Lee, YE, Park, CW, Hong, SY, and Na, SI, 2016, Study on Reflectance and NDVI of Aerial Images using a Fixed-Wing UAV eBee, Korean Journal of Soil Science and Fertilizer 49(6) , pp.731-742. Mireia G, Marta C, David M, Payam D, Antonia V, Antoni P, and Mark JN., 2016, Normalized difference vegetation index (NDVI) as a marker of surrounding greenness in epidemiological studies: The case of Barcelona city, Urban Forestry & Urban Greening 19, pp.88-94. Peng, J., Hu, Y., Dong, J., Liu, Q., Liu, Y., 2020, Quantifying spatial morphology and connectivity of urban heat islands in a megacity: A radius approach, Sci Total Environ 714, doi :10.1016/j.scitotenv.2020.136792. Ranjay S, Liping D, Eugene GY, Lingjun K, Yuan-zheng S, and Yu-qi B., 2017, Regression model to estimate flood impact on corn yield using MODIS NDVI and USDA cropland data layer, Journal of Integrative Agriculture 16, pp.398-407. Rouse, J., Haas, RH, Schell, JA, 1974, Monitoring vegetation systems in the Great Plains with ERTS 351, p.309. Sabrin, S., Karimi, M., Nazari, R., 2020, Developing Vulnerability Index to Quantify Urban Heat Islands Effects Coupled with Air Pollution: A Case Study of Camden, International Journal of Geo-Information 9, doi:ARTN 34910.3390/ ijgi9060349. Shrestha, R., Di, LP, Yu, EG, Kang, LJ, Shao, YZ, Bai, YQ, 2017, Regression model to estimate flood impact on corn yield using MODIS NDVI and USDA cropland data layer, J Integr Agr 16, pp.398-407. Sotille, ME, Bremer, UF, Vieira, G., Velho, LF, Petsch, C., Simoes, JC, 2020, Evaluation of UAV and satellite-derived NDVI to map maritime Antarctic vegetation, Applied Geography 125, doi:ARTN 10232210.1016 /j.apgeog.102322. Su, YX, Liu, LY, Wu, JP, Chen, on local air temperature using a novel three-layered land surface energy balance model, Environment International 132, doi:ARTN 10508010.1016/j.envint.2019.105080. Tomas, JR and HW Gausman, 1977, Leat reflectance vs. leaf chlorophyll and carotenoid concentrations for eight crops, Agronomy Journal 69, pp.799-802. Uddin, MN, Islam, AKMS, Bala, SK, Islam, GMT, Adhikary, S., Saha, D., Haque, S., Fahad, MGR, Akter, R., 2019, Mapping of climate vulnerability of the coastal region of Bangladesh using principal component analysis, Applied Geography 102, pp.47-57. Voogt, JA, Oke, TR, 2003, Thermal remote sensing of urban climates, Remote Sensing of Environment 86, pp.370-384. Xingwang F. and Yuanbo L., 2016, A global study of NDVI differences among moderate-resolution satellite sensors. Journal of Photogrammetry and Remote Sensing 121, pp.177-191. Zhang Feng, Zhang Li-Wen, Shi Jing-Jing, and Huang Jing-Feng, 2014, Soil Moisture Monitoring Based on Land Surface Temperature Vegetation Index Space derived from MODIS Data. PEDOSPHERE 24(4):450-460. Zhao, F., Wu, X., Wang, SJ, Object-oriented Vegetation Classification Method based on UAV and Satellite Image Fusion 174, pp.609-615. Zhou, Electron Agr 168, doi:ARTN 10514410.1016/j.compag.2019.105144. Ziter, CD, Pedersen, EJ, Kucharik, CJ, Turner, MG, 2019, Scale-dependent interactions between tree canopy cover and impervious surfaces reduce daytime urban heat during summer, Proceedings of the National Academy of Sciences of the United States of America, 116, pp.7575-7580.

The purpose of the present invention to solve the above problems is to accurately analyze the vegetation distribution area reflecting the urban cover characteristics, to solve the problem of misclassification of vegetation in urban areas in the areas of steel plate roofs, waterproof coating roofs, and urethane coating areas, and to accurately analyze vegetation distribution areas reflecting urban cover characteristics. By developing new vegetation index calculation formulas (1), (2), (3) to classify accurate vegetation areas by considering the cover characteristics of urban areas, the vegetation of urban areas with land cover is analyzed, taking into account the land cover characteristics of urban areas. We provide a customized vegetation index measurement method based on UAV multi-spectral sensors.

In order to achieve the purpose of the present invention, a customized vegetation index measurement method based on a UAV multi-spectral sensor considering the land cover characteristics of urban areas is (a) a camera of a drone for the target area for urban area vegetation analysis by urban area land cover class. To match the coordinates of the image, GNSS surveying is performed to select k ground control points (GCP), and optical (RGB) and near-infrared (Nir) images are captured through video shooting by a multi-spectral sensor (camera) mounted on a drone flying at a certain flight altitude. ) Acquiring a multispectral image including; (b) combining the images using image registration software, thereby constructing orthoimages and multispectral image index maps for Red, Green, Blue, and Near-IR bands; and (c) existing NDVI, NDVI, and GNDVI from images for each band of Red, Green, Blue, and Near-IR bands using GIS SW to compare and analyze vegetation distribution areas reflecting urban cover characteristics by cover of the target area. , BNDVI, RGBVI, GRVI, and SAVI vegetation indices are calculated, and the vegetation index is calculated by applying the new vegetation index calculation formula (3) to analyze the vegetation distribution area reflecting urban cover characteristics, and the accuracy of vegetation classification using the Kappa coefficient is compared. It includes the step of classifying steel plate roofs, waterproof coating roofs, and urethane coating areas without misclassification, and comparing and analyzing existing vegetation indices and urban cover characteristics,
The new vegetation index calculation formula (3) uses Blue, Red, and Nir values to It is calculated and accurately classifies vegetation considering the land cover characteristics of urban areas without misclassification of steel plate roofs, waterproof coating roofs, and urethane coating areas.

The customized vegetation index measurement method based on a UAV multi-spectral sensor considering the land cover characteristics of urban areas of the present invention classifies accurate vegetation areas by considering various cover characteristics found in urban areas, such as steel plate roofs, waterproof coated roofs, and urethane coated areas. A new vegetation index (calculation 3) was developed.

Satellite images have been widely used in the fields of urban planning, landscape management, and environmental monitoring, but have many disadvantages such as difficulty in rapid shooting and high cost. Recently, image capture technology using unmanned aerial vehicles (UAVs) equipped with multispectral cameras has been developed. It is being introduced. In particular, interest in urban regeneration projects is increasing in Korea, and UAV-based vegetation analysis results in urban areas are receiving attention as important data for effectively promoting urban regeneration projects. Previously, vegetation information over a wide area was acquired using multi-spectral bands of satellites, but recently, UAV technology that can obtain vegetation information in a faster and more accurate manner is being actively researched. The present invention acquires multispectral images using a UAV equipped with a Micasense RedEde MX camera, and measures NDVI (Normalized Difference Vegetation Index), GNDVI (Green Normalized Difference Vegetation Index), BNDVI (Blue Normalized Dif-ference Vegetation Index), and RGBVI ( Vegetation indices such as Red Green Blue Vegetation Index (GRVI), Green Red Vegetation Index (GRVI) and Soil Adjusted Vegetation Index (SAVI) were analyzed. However, in the process of analyzing vegetation in urban areas using existing vegetation indices, there were cases where the vegetation index values of steel plate roofs, waterproof coated roofs, and urethane coated areas were similar to or slightly higher than those of vegetation areas such as grass. . To improve this vegetation misclassification problem, various types of vegetation index equations were developed using multi-spectral images captured by UAV. As a result of performing Kappa coefficient analysis on the developed vegetation index, the squared Red-Blue NDVI index showed the best results when analyzing vegetation that reflects the land cover of urban areas. The new vegetation index developed in this study will be very useful in effectively analyzing the vegetation of urban areas with various types of land cover, such as steel plate roofs, waterproof coating roofs, and basketball courts constructed with urethane coating.

The new vegetation index calculation formula (3) showed the highest Kappa coefficient of 1.0. Therefore, when applying the vegetation index calculation formula (3), it was possible to most effectively classify vegetation without misclassification in urban areas with various land cover characteristics such as steel sheet roofs, waterproof coating roofs, and urethane coatings.

In future urban regeneration projects, urban planning, landscape and environmental work, the vegetation index calculation formula (3) newly developed in this study will be used as a standard technology to accurately classify vegetation from satellite images and unmanned aerial vehicle (UAV) images. will be.

Figure 1 is a flow chart showing a customized vegetation index measurement method based on a UAV multi-spectral sensor considering the land cover characteristics of urban areas according to the present invention.
Figure 2 is a photograph of a study site for urban vegetation analysis considering the land cover characteristics of urban areas.
Figure 3 is a photo of a rotary wing drone (DJI Matrice 210 drone) equipped with a Micasense RedEdge MX multi-spectral sensor.
Figure 4 is a photo of establishing a UAV (drone) flight plan using the drone's flight planning program (DJI GS Pro program).
Figure 5 is a UAV image processing screen using image registration software (Pix4D Mapper S/W).
Figure 6 is a distribution chart of the existing NDVI, NDVI, GNDVI, BNDVI, RGBVI, GRVI, and SAVI vegetation indices for the target area in order to analyze the vegetation distribution area reflecting the urban cover characteristics.
Figure 7 is a diagram showing the selection of sample points by land cover.
Figure 8 shows the vegetation index distribution map and vegetation classification results analyzed by applying the vegetation index calculation formula (1).
Figure 9 shows the vegetation index distribution map and vegetation classification results analyzed by applying the vegetation index calculation formula (2).
Figure 10 shows the vegetation index distribution and vegetation classification results analyzed by applying the vegetation index calculation formula (3) according to the present invention.
Figure 11 shows the verification points for each land cover type in the vegetation map created by applying the new vegetation index calculation formulas (1), (2), and (3), respectively.

Hereinafter, the configuration and operation of the preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the description of the present invention, if it is determined that a detailed description of a related known function or configuration may unnecessarily obscure the gist of the present invention, the detailed description will be omitted. In addition, the same drawing number is assigned to different drawings when indicating the same configuration.

This study developed a new vegetation index (calculation 3) to accurately classify vegetation areas by considering the various cover characteristics found in urban areas, such as steel sheet roofs, waterproof coating roofs, and urethane coating areas.

delete

Figure 1 is a flow chart showing a customized vegetation index measurement method based on a UAV multi-spectral sensor considering the land cover characteristics of urban areas according to the present invention.

A UAV multi-spectral sensor-based customized vegetation index measurement method considering the land cover characteristics of urban areas using a drone camera (Micasense RedEdge MX multi-spectral sensor) is

(a) Considering land cover characteristics for each urban area land cover class, k ground control points (GCP) are selected by performing GNSS surveying to match the coordinates of the drone camera image for the target area for urban area vegetation analysis, After establishing a drone flight plan using the drone's flight plan software (DJI GS Pro S/W), video is captured by the Micasense RedEdge MX multi-spectral sensor (drone camera) mounted on the drone flying at a flight altitude of 100m. Acquiring multispectral images including optical (RGB) and near-infrared (Nir);

(b) Images are combined using image registration software (Pix4D Mapper S/W), through which orthoimages for Red, Green, Blue, and Near-IR bands are constructed, and photos, GPS, INS (Inertial Navigation System, Constructing an orthoimage and multispectral image index map generated through image splicing technology for single photos taken from (inertial navigation device) information; and

(c) In order to compare and analyze the vegetation distribution area reflecting the urban cover characteristics by cover of the target area, using GIS SW, the existing NDVI, NDVI, GNDVI, BNDVI, RGBVI, GRVI, and SAVI vegetation indices are calculated, and the vegetation index is calculated by applying the new vegetation index calculation formula (3) to analyze the vegetation distribution area reflecting urban cover characteristics, and the accuracy of vegetation classification using the Kappa coefficient is compared. It includes the step of classifying steel plate roofs, waterproof coating roofs, and urethane coating areas without misclassification, and comparing and analyzing existing vegetation indices (NDVI, NDVI, GNDVI, BNDVI, RGBVI, GRVI, SAVI vegetation indices) and urban cover characteristics,
The new vegetation index calculation formula (3) uses Blue, Red, and Nir values to It is calculated and accurately classifies vegetation considering the land cover characteristics of urban areas without misclassification of steel plate roofs, waterproof coating roofs, and urethane coating areas.

The drone may be a fixed-wing drone or a rotary-wing drone.

A rotary-wing drone has 4, 6, or 8 propellers, a motor connected to each propeller, an electronic speed controller (ESC), a flight controller (FC), a GNSS receiver, a gyroscope, and an acceleration sensor. It is equipped with an INS (Inertial Navigation System), an altimeter, and a battery.

In an embodiment, the drone is equipped with an INS (Inertial Navigation System) including a GNSS receiver, an altimeter, a gyroscope, and an acceleration sensor, and acquires Blue, Green, Red, Red edge, and Near-IR spectroscopic images. A rotary wing drone equipped with a Micasense RedEdge MX multi-spectral sensor is used, and the Micasense RedEdge MX multi-spectral sensor (camera) can acquire Blue, Green, Red, Red edge, and Near-IR spectral images. Therefore, it is used in vegetation analysis, flood monitoring, and water quality analysis research.

The urban area land cover class includes steel sheet roofs, artificial turf, playground tracks, urethane coated areas, waterproof coated roofs, lawns, trees, and buildings.

The process of conducting this study is shown in Figure 1 for the customized vegetation index measurement method based on UAV multispectral sensors that considers land cover characteristics in urban areas using a drone (UAV) camera. First, as shown in Figure 2, urban area vegetation is considered by considering land cover characteristics for each urban area land cover class (steel roof, artificial turf, playground track, urethane coated area, waterproof coated roof, grass, trees, and building). A research site was selected for analysis, and k ground control points (GCPs) were selected by performing GNSS (Global Navigation Satellite System) surveying to match the coordinates of drone images. After mounting the drone's camera (Micasense RedEdge MX multi-spectral sensor) on the drone, a drone flight plan was established using the drone's flight plan software (DJI GS Pro S/W), and optical ( Multispectral images such as RGB) and near infrared (NIR) were acquired. In addition, the images were combined using image registration software (Pix4D Mapper S/W), and through this, orthoimages for the Red, Green, Blue, and Near-IR bands were constructed. And, using GIS SW (ArcGIS S/W), various vegetation indices such as existing NDVI, NDVI, GNDVI, BNDVI, RGBVI, GRVI, and SAVI were calculated from the images for each band. By comparing and analyzing the vegetation index values for each type of cover that makes up the target area, points showing abnormal values were identified, and aspects to be considered when analyzing vegetation in urban areas were reviewed. And finally, new vegetation index calculation formulas (1), (2), and (3) were derived and applied to effectively analyze the vegetation distribution area reflecting the urban cover characteristics.

Satellite images have been widely used in the fields of urban planning, landscape management, and environmental monitoring, but have many disadvantages such as difficulty in rapid shooting and high cost. Recently, image capture technology using unmanned aerial vehicles (UAVs) equipped with multispectral cameras has been developed. It is being introduced. In particular, interest in urban regeneration projects is increasing in Korea, and UAV-based vegetation analysis results in urban areas are receiving attention as important data for effectively promoting urban regeneration projects. Previously, vegetation information over a wide area was acquired using multi-spectral bands of satellites, but recently, UAV technology that can obtain vegetation information in a faster and more accurate manner is being actively researched. The present invention acquires multispectral images using a UAV equipped with a Micasense RedEde MX camera, and measures NDVI (Normalized Difference Vegetation Index), GNDVI (Green Normalized Difference Vegetation Index), BNDVI (Blue Normalized Dif-ference Vegetation Index), and RGBVI ( Vegetation indices such as Red Green Blue Vegetation Index (GRVI), Green Red Vegetation Index (GRVI) and Soil Adjusted Vegetation Index (SAVI) were analyzed. However, in the process of analyzing vegetation in urban areas using existing vegetation indices, there were cases where the vegetation index values of steel plate roofs, waterproof coated roofs, and urethane coated areas were similar to or slightly higher than those of vegetation areas such as grass. . To improve this vegetation misclassification problem, various types of vegetation index equations were developed using multispectral images captured by UAV. As a result of performing Kappa coefficient analysis on the developed vegetation index, the squared Red-Blue NDVI index showed the best results when analyzing vegetation that reflects the land cover of urban areas. The new vegetation index developed in this study will be very useful in effectively classifying vegetation in urban areas with various types of land cover, such as steel sheet roofs, waterproof coating roofs, and basketball courts constructed with urethane coating.

2. Target site selection and drone image processing

1) Research site

Figure 2 is a photograph of the study site for urban area vegetation analysis considering the land cover characteristics of the urban area. As shown in Figure 2, this study selected some areas of ○○ University, located in Jeonju, Jeollabuk-do, Korea, as a research site for analysis of vegetation in urban areas. The area consists of a variety of coverings, including vegetated areas such as trees and grass, as well as artificial turf, playground tracks, urethane basketball courts, steel sheet and waterproof coated roofs, and non-vegetated areas.

2) Drone video shooting and video splicing processing

In this study, a Micasense RedEdge MX multispectral sensor (camera) was mounted on a DJI Matrice 210 drone (UAV) as shown in Figure 3 to obtain multispectral images of the target area for vegetation analysis in urban areas. A drone (UAV) uses a fixed-wing drone or a rotary-wing drone, and in the embodiment, a rotary-wing drone equipped with a Micasense RedEdge MX multi-spectral sensor, a camera capable of acquiring blue, green, red, red edge, and near-IR spectral images. used.

Micasense RedEdge MX multi-spectral sensor (camera) acquires Blue, Green, Red, Red edge (longer wavelength than Red of the Red outline and smaller wavelength than Near-IR), and Near-IR spectral images as shown in Table 1. Because it can do so, it is widely used in research such as vegetation analysis, flood monitoring, and water quality analysis. The resolution is about 8 cm at an altitude of 120 m, providing very precise image information, and the weight is very light at 231.9 g, so it has the advantage of securing sufficient flight time for the DJI Matrice 210 drone.

Figure 3 is a photo of a rotary wing drone (DJI Matrice 210 drone) equipped with a Micasense RedEdge MX multi-spectral sensor. Figure 4 is a photo of establishing a UAV (drone) flight plan using the drone's flight plan program (DJI GS Pro program).

The drone may be a fixed-wing drone or a rotary-wing drone.

A rotary-wing drone has 4, 6, or 8 propellers, a motor connected to each propeller, an electronic speed controller (ESC), a flight controller (FC), a GNSS receiver, a gyroscope, and an acceleration sensor. It is equipped with an INS (Inertial Navigation System), an altimeter, and a battery.

To establish a drone flight plan, drone flight plan SW (DJI GS Pro program) was used as shown in Figure 4. Vertical overlap and lateral overlap were designed to be 80% and 70%, respectively, and drone camera shooting was performed at a flight altitude of about 100 m to acquire multispectral images from a camera (Micasense RedEdge MX sensor) with a resolution of 6 to 7 cm. did.

Table 1 shows Micasense RedEdge MX sensor specifications.

To ensure the location accuracy of the drone image, 8 ground reference points were selected as shown in Table 2, and the TM (Transverse Mercator) coordinates of the GRS80 ellipsoid were acquired through GNSS surveying. The captured video has UTM (Universal Transverse Mercator) coordinates of the WGS84 ellipsoid through the GNSS receiver mounted on the drone. Therefore, in order to convert to TM coordinates of the GRS80 ellipsoid used in Korea, matching with the survey results of 8 ground control points was performed. And, as shown in Figure 5, an orthoimage was created by splicing the images using image registration software (Pix4D Mapper S/W), and to calculate the vegetation index, indexes were collected for each red, green, blue, and near-infrared (Nir) multispectral image. A map was built.

Table 2 shows the ground control point (GCP) coordinates.

X(E), Y(N) refer to the TM (Transverse Mercator) coordinates based on the GRS80 ellipsoid based on the central origin, 127°E, 38°N. In other words, it is a value expressed in meters (m) along the horizontal axis (East or However, in Korea, if 127°E 38°N is used as the origin, a negative number (-) may occur, so the actual origin is added by +200,000m and +600,000m on the horizontal and vertical axes, respectively, from 127°E 38°N. I am using it. Z(EL.m) refers to the height based on mean sea level, that is, elevation or elevation value above sea level.

Figure 5 is a UAV image processing screen using image registration software (Pix4D Mapper S/W).

3. Development of vegetation analysis technology reflecting urban cover characteristics

1) Vegetation analysis using existing vegetation indices (NDVI, NDVI, GNDVI, BNDVI, RGBVI, GRVI, SAVI)

This study analyzed the vegetation index as shown in Table 3 using ArcGIS S/W based on Blue, Green, Red, and Near-IR images constructed through image processing of a drone camera.

Figure 6 is a distribution chart of the existing NDVI, NDVI, GNDVI, BNDVI, RGBVI, GRVI, and SAVI vegetation indices for the target area in order to analyze the vegetation distribution area reflecting the urban cover characteristics, Table 3 is the vegetation index calculation formula, and Table 4 is the vegetation index. It shows the results of analyzing the statistical characteristics of each index.

Table 4 shows the statistical characteristics of each existing NDVI, NDVI, GNDVI, BNDVI, RGBVI, GRVI, and SAVI vegetation index.

Sites in urban areas are composed of various coverings such as trees, grass, artificial turf, playground tracks, urethane basketball courts, steel sheets, and waterproof coated roofs.

The urban land cover class includes steel sheet roofs, artificial turf, playground tracks, urethane coated areas, waterproof coated roofs, lawns, trees, and buildings.

Figure 7 is a diagram showing the selection of sample points by land cover.

Therefore, as shown in Figure 7, locations that could represent each land cover were selected and 3 sample points were selected for each location, with 2 each for grass (No. 19 to 24) and trees (No. 25 to 30). The location of the dog was selected. April 2, 2021, when UAV filming was conducted, corresponds to early spring season, and the vegetation vitality of evergreen trees is relatively high at this time. However, since “grass” usually grows a lot after May, it shows slightly lower vegetation vitality than “trees” on April 2. In this study, locations with normal vegetation vitality (No. 19 to 21) and locations with slightly low vegetation vitality (No. 22 to 24) were selected and compared.

Table 5 shows data from a comparative analysis of six vegetation index values by land cover using the existing vegetation index.

① NDVI vegetation index analysis

The analysis results of the NDVI vegetation index are as follows. The vegetation index of the steel plate roof (No. 1~3) was 0.48~0.53, and the vegetation vitality was similar to the vegetation index of 0.50~0.56 of the grass (No. 19~21) with normal vegetation vitality. It was higher than the slightly lower vegetation index of grass (No. 22-24), which was 0.24-0.35. In addition, the vegetation index of the green urethane coating (No.10~12) and green waterproof coating (No.16~18) was 0.24~0.32 and 0.30~0.35, respectively, indicating that the vegetation index was slightly lower than that of grass (No.22~). 24) was found to be similar to the vegetation index of 0.24 to 0.35.

Therefore, when vegetation is classified by setting the vegetation index of grass with average vegetation vitality as the threshold, the steel sheet roof is misclassified as vegetation. In addition, when the vegetation index of grass with slightly low vegetation vitality is set as the threshold to classify vegetation, the steel sheet roof, green urethane coating, and green waterproof coating areas result in misclassification as vegetation. do.

② GNDVI vegetation index analysis

The analysis results of the GNDVI vegetation index are as follows. The vegetation indices of the steel plate roof (No.1~3) and the red urethane coating (No.13~15) were 0.38~0.44 and 0.39~0.41, respectively, compared to the grass (No.19~21) with normal vegetation vitality. It was similar to the vegetation index of 0.41 to 0.46, and was greater than the vegetation index of 0.27 to 0.33 for grass (No. 22 to 24), which had slightly lower vegetation vitality.

Therefore, when vegetation is classified by setting the vegetation index of grass with normal or slightly low vegetation vitality as the threshold, the steel sheet roof and the red urethane coating area result in misclassification as vegetation.

③ BNDVI vegetation index analysis

The analysis results of the BNDVI vegetation index are as follows. The vegetation index of red urethane coating (No.13~15) was 0.6, which was lower than that of grass (No.19~21) with normal vegetation vitality, but the vegetation index of grass (No.19~21) with slightly lower vegetation vitality was 0.6. 21) was similar to the vegetation index of 0.42 to 0.50.

Therefore, if you classify vegetation by setting the vegetation index of grass with normal vegetation vitality as the threshold, you can analyze the vegetation area relatively effectively, but you can analyze the vegetation area relatively effectively, but you can use the vegetation index of grass with slightly low vegetation vitality as the threshold ( If you classify vegetation by setting the threshold, the red urethane coating area will be misclassified as vegetation.

④ RGBVI vegetation index analysis

The analysis results of the RGBVI vegetation index are as follows. The vegetation index of green urethane coating (No.10~12) and green waterproof coating (No.16~18) is 0.34~0.40 and 0.30~0.31, respectively, compared to grass (No.19~21) with normal vegetation vitality. It was similar to the vegetation index of 0.29 to 0.37, and was greater than the vegetation index of 0.13 to 0.24 for grass (No. 22 to 24), which had slightly lower vegetation vitality.

Therefore, when vegetation is classified by setting the vegetation index of grass with normal or slightly low vegetation vitality as the threshold, the green urethane coating and green waterproof coating areas are misclassified as vegetation.

⑤ GRVI vegetation index analysis

The analysis results of the GRVI vegetation index are as follows. The vegetation index of the steel plate roof (No. 1~3) was 0.11~0.12, which was similar to the vegetation index of 0.09~0.14 for the grass (No. 19~21), which had average vegetation vitality. It was higher than the slightly lower vegetation index of grass (No. 22~24), -0.03~0.03. In addition, the vegetation index of green urethane coating (No.10~12) and green waterproof coating (No.16~18) is 0.22~0.27 and 0.21~0.22, respectively, which is equivalent to grass with normal vegetation vitality (No.19~ 21), which was higher than the vegetation index of 0.09 to 0.14.

Therefore, when vegetation is classified by setting the vegetation index of grass with normal or slightly low vegetation vitality as the threshold, the steel sheet roof, green urethane coating, and green waterproof coating areas are misclassified as vegetation. It will be brought.

⑥ SAVI vegetation index analysis

The analysis results of the SAVI vegetation index are as follows. The vegetation index of the steel plate roof (No. 1~3) was 0.72~0.79, which was similar to the vegetation index of 0.74~0.84 for the grass (No. 19~21) with normal vegetation vitality, and the vegetation vitality was slightly It was greater than the vegetation index of low grass (No. 22~24), which was 0.36~0.53. In addition, the vegetation index of green urethane coating (No.10~12) and green waterproof coating (No.16~18) is 0.37~0.48 and 0.44~0.53, respectively, which indicates that grass (No.22~24) has slightly lower vegetation vitality. ) was found to be similar to the vegetation index of 0.36 to 0.53.

Therefore, if vegetation is classified by setting the vegetation index of grass with average vegetation vitality as the threshold, the steel sheet roof will be misclassified as vegetation. In addition, if vegetation is classified by setting the vegetation index of grass with slightly low vegetation vitality as the threshold, the steel sheet roof, green urethane coating, and green waterproof coating areas are misclassified as vegetation. . Therefore, the SAVI vegetation index, which reflects urban land cover characteristics, showed a similar pattern to the NDVI vegetation index.

2) Development of a vegetation index based on UAV multi-spectral sensors reflecting urban area cover characteristics

In urban areas, when using the existing six vegetation indices such as NDVI, NDVI, GNDVI, BNDVI, RGBVI, GRVI, and SAVI, it was confirmed that a problem occurred in which steel plate roofs, urethane coatings, and waterproof coating areas were misclassified as vegetation.

In order to improve this problem, this study tested various types of vegetation indices using Blue, Green, Red, and Near-IR bands, and among these, as shown in Table 6, three final vegetation index calculation formulas were developed to calculate the It was used to classify vegetation. Table 6 shows the new vegetation index calculation formulas (1), (2), and (3) that classify vegetation in urban areas.

Table 7 shows the results of comparative analysis using the new vegetation index calculation formulas (1), (2), and (3) that classify vegetation in urban areas. In addition, Figures 8 to 11 show the vegetation index distribution map analyzed using the new vegetation index calculation formulas (1), (2), and (3) and the results of vegetation classification according to boundary values.

First, when the vegetation index calculation formula (1) is applied, the vegetation index of the red urethane coating (No. 13 to 15) is 0.72 to 0.74, which is 0.62, which is the vegetation index of grass (No. 22 to 24) with slightly lower vegetation vitality. It was slightly higher than ~0.71. Therefore, as shown in Figure 8(b), the red urethane coating (No. 13 to 15) area was misclassified as vegetation.

In addition, when the vegetation index calculation formula (2) is applied, the vegetation index of the steel plate roof (No. 1 to 3) is 0.73 to 0.78, which is 0.76 to 0.76, which is the vegetation index of grass (No. 19 to 21) with normal vegetation vitality. It was similar to 0.81, and was higher than the vegetation index of 0.48 to 0.61 for grass (No. 22 to 24), which has slightly lower vegetation vitality. Therefore, as shown in Figure 9(b), the area of the steel plate roof (No. 1 to 3) was misclassified as vegetation.

The new vegetation index calculation formula (3) in urban areas according to the present invention uses Blue, Red, and Nir values. It is calculated as

In addition, when applying the new vegetation index calculation formula (3), the vegetation index of steel sheet roof (No.1~3), urethane coating (No.10~15), and waterproof coating (No.16~18) is the vegetation vitality. All were lower than the slightly lower vegetation index of grass (No. 22 to 24), which was 0.60 to 0.73. Therefore, as shown in Figure 10(b), the steel plate roof (No. 1~3), waterproof coating roof (No. 16~18), and urethane coating (No. 10~15) areas are all non-vegetation areas. were classified, and only tree and grass areas were accurately classified as vegetation areas.

Table 7 shows comparison data of vegetation index values by land cover using the new vegetation index.

Figure 8 shows the vegetation index distribution chart and vegetation classification results analyzed by applying the vegetation index calculation formula (1). Figure 9 shows the vegetation index distribution map and vegetation classification results analyzed by applying the vegetation index calculation formula (2). Figure 10 shows the vegetation index distribution map and vegetation classification results analyzed by applying the vegetation index calculation formula (3).

The new vegetation index calculation formula (3) uses Blue, Red, and Nir values to Calculated as , vegetation was accurately classified considering the land cover characteristics of urban areas without misclassification of steel plate roof, waterproof coating roof, and urethane coating area.

In order to quantitatively verify the three vegetation index calculation formulas newly proposed in this study, five verification points were selected for each land cover type through field surveys. Among these, 20 verification points with trees and lawns (including grass) were designated as vegetation areas, and 30 verification points with steel plate roofs, waterproof coating roofs, artificial turf, playground tracks, and urethane coating areas were designated as vegetation areas. It was designated as a non-vegetation area.

Figure 11 shows the verification points for each land cover type in the vegetation map created by applying the new vegetation index calculation formulas (1), (2), and (3), respectively.

Table 8 shows the results of analysis by applying the new vegetation map calculation formula and the results of comparison with the actual survey results in terms of Kappa coefficient. The closer the Kappa coefficient is to 1, the higher the accuracy. The Kappa coefficients of vegetation index calculation formulas (1) and (2) were both 0.8, and vegetation index calculation formula (3) showed the highest Kappa coefficient of 1.0. As a result, when applying the vegetation index calculation formula (3), it was possible to most effectively classify vegetation without misclassification in urban areas with various land cover characteristics such as steel plate roofs, waterproof coating roofs, and urethane coatings.

In the future, in promoting urban regeneration projects, urban planning, landscape, and environmental work, the newly developed vegetation index calculation formula (3) in this study will be established as a standard technology for accurately classifying vegetation from satellite and unmanned aerial vehicle (UAV) images. It is judged that it will be done. Table 8 shows a comparison of vegetation classification accuracy using the Kappa coefficient.

4. Conclusion

Obtaining accurate vegetation information is very important in various fields such as urban planning, landscaping, water resources and water quality analysis. Recently, research is being conducted to quantitatively analyze urban vegetation distribution using satellite images (landsat images) or unmanned aerial vehicle (UAV) images to effectively promote urban regeneration projects. In particular, vegetation indices such as NDVI are widely used in vegetation analysis using remote sensing technologies such as satellite images and unmanned aerial vehicle (UAV) images.

In urban areas, there are many different types of land cover, such as steel sheet roofs, waterproof coated roofs, and urethane coated areas. In this study, the formulas for calculating existing vegetation indices such as NDVI are similar to steel sheet roofs, waterproof coated roofs, and urethane coated areas. There was a problem that the vegetation index value of the coating area misclassified vegetation in urban areas with land cover such as grass. To examine these issues, we first quantitatively analyzed vegetation index values for various land cover types using the existing six vegetation indices, including NDVI, GNDVI, BNDVI, RGB-VI, GRVI, and SAVI.

For this purpose, we compared and analyzed various types of vegetation indices such as NDVI using high-resolution multispectral images taken at a flight altitude of 100m by a drone (UAV) equipped with a Micasense RedEde MX multispectral sensor (camera). As a result, the vegetation index values of the steel plate roof, waterproof coating roof, and urethane coating areas were misclassified as vegetation such as grass.

To solve the problem of misclassification of vegetation in steel plate roofs, waterproof coating roofs, and urethane coating areas according to urban land cover characteristics, multispectral images were combined by RGB and NIR bands to calculate three vegetation indices (1), (2), (3) was derived, and the accuracy of the three newly proposed vegetation index calculation formulas was evaluated using the Kappa coefficient.

As a result, the new vegetation index calculation formula (3) was created by combining blue, red, and near-infrared (Near-IR). ) was found to classify vegetation most accurately reflecting the land cover characteristics of urban areas. Therefore, the new vegetation index presented in this study is expected to be very useful in classifying vegetation in urban areas, including future urban regeneration projects.

Embodiments according to the present invention are implemented in the form of program instructions that can be executed through various computer means and can be recorded on a computer-readable recording medium. The computer-readable recording medium may include program instructions, data files, and data structures alone or in combination. Computer-readable recording media include storage, magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. -Hardware devices configured to store and execute program instructions in storage media such as magneto-optical media, ROM, RAM, flash memory, etc. may be included. Examples of program instructions may include those produced by a compiler, machine language code, as well as high-level language code that can be executed by a computer using an interpreter. The hardware device may be configured to operate as one or more software modules to perform the operations of the present invention.

As described above, the method of the present invention is implemented as a program and can be stored on a recording medium (CD-ROM, RAM, ROM, memory card, hard disk, magneto-optical disk, storage device, etc.) in a form that can be read using computer software. ) can be stored in .

Although the present invention has been described with reference to specific embodiments of the present invention, the present invention is not limited to the same configuration and operation as the specific embodiments to illustrate the technical idea as described above, and is not limited to the technical idea and scope of the present invention. It can be implemented with various modifications, and the scope of the present invention should be determined by the claims described later.

Claims (7)

(a) For each urban area land cover class, k ground control points (GCP) are selected by performing GNSS surveying to match the coordinates of the drone's camera image for the target area for urban area vegetation analysis, and the drone flies at a certain flight altitude. Acquiring multi-spectral images including optical (RGB) and near-infrared (Nir) rays through image capture by a multi-spectral sensor (camera) mounted on the .
(b) combining the images using image registration software, thereby constructing orthoimages and multispectral image index maps for Red, Green, Blue, and Near-IR bands; and
(c) In order to compare and analyze the vegetation distribution area reflecting the urban cover characteristics by cover of the target area, using GIS SW, the existing NDVI, NDVI, GNDVI, Calculate the BNDVI, RGBVI, GRVI, and SAVI vegetation indices, and apply the new vegetation index calculation formula (3) to analyze vegetation distribution areas that reflect urban cover characteristics to calculate the vegetation index and compare the accuracy of vegetation classification using the Kappa coefficient. It includes the step of classifying steel plate roofs, waterproof coating roofs, and urethane coating areas without misclassification, and comparing and analyzing existing vegetation indices and urban cover characteristics,
The new vegetation index calculation formula (3) uses Blue, Red, and Nir values to It is calculated as a customized vegetation index measurement based on a UAV multi-spectral sensor considering the land cover characteristics of urban areas, which accurately classifies vegetation considering the land cover characteristics of urban areas without misclassification of steel plate roofs, waterproof coating roofs, and urethane coating areas. method. According to paragraph 1,
The drone is a customized vegetation index measurement method based on a UAV multi-spectral sensor that considers land cover characteristics in urban areas using a fixed-wing drone or rotary-wing drone. According to paragraph 2,
The rotary wing drone is equipped with an INS (Inertial Navigation System) equipped with a GNSS receiver, altimeter, gyroscope, and acceleration sensor, and Micasense RedEdge, a camera that can acquire Blue, Green, Red, Red edge, and Near-IR spectral images. A rotary-wing drone equipped with an MX multi-spectral sensor is used, and the Micasense RedEdge MX multi-spectral sensor (camera) can acquire Blue, Green, Red, Red edge, and Near-IR spectral images for vegetation analysis, flood monitoring, and a customized vegetation index measurement method based on UAV multispectral sensors that consider land cover characteristics in urban areas, used in water quality analysis research. According to paragraph 1,
The urban land cover class includes customized vegetation based on UAV multi-spectral sensors considering the land cover characteristics of urban areas, including steel plate roofs, artificial turf, playground tracks, urethane coated areas, waterproof coated roofs, grass, trees, and buildings. How to measure exponents. delete According to paragraph 1,
When applying the new vegetation index calculation formula (3) above, the vegetation index of steel sheet roof (No.1~3), urethane coating (No.10~15), and waterproof coating (No.16~18) showed slightly lower vegetation vitality. All were lower than the vegetation index of 0.60 to 0.73 for low grass (No. 22 to 24), and steel sheet roof (No. 1 to 3), waterproof coating roof (No. 16 to 18), and urethane coating (No. 10~15) A customized vegetation index measurement method based on UAV multi-spectral sensors that takes into account the land cover characteristics of urban areas, where all areas are classified as non-vegetation areas and only tree and grass areas are accurately classified as vegetation areas. According to paragraph 1,
The new vegetation index calculation formula (3) is used as a technology to accurately classify urban area vegetation from satellite images and unmanned aerial vehicle (UAV) images when promoting urban regeneration projects, urban planning, landscape, and environmental work. Customized vegetation index measurement method based on UAV multispectral sensor considering regional land cover characteristics.