A half-Gaussian fitting method for estimating fractional vegetation cover of corn crops using unmanned aerial vehicle images

Linyuan Li, Xihan Mu, Craig Macfarlane, Wanjuan Song, Jun Chen, Kai Yan, Guangjian Yan

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

Accurate estimates of fractional vegetation cover (FVC) using remotely sensed images collected using unmanned aerial vehicles (UAVs) offer considerable potential for field measurement. However, most existing methods, which were originally designed to extract FVC from ground-based remotely sensed images (acquired at a few meters above the ground level), cannot be directly used to process aerial images because of the presence of large quantities of mixed pixels. To alleviate the negative effects of mixed pixels, we proposed a new method for decomposing the Gaussian mixture model and estimating FVC, namely, the half-Gaussian fitting method for FVC estimation (HAGFVC). In this method, the histograms of pure vegetation pixels and pure background pixels are firstly fit using two half-Gaussian distributions in the Commission Internationale d'Eclairage (CIE) L*a*b* color space. Then, a threshold is determined based on the parameters of Gaussian distribution to generate a more accurate FVC estimate. We acquired low-altitude remote-sensing (LARS) images in three vegetative growth stages at different flight altitudes over a cornfield. The HAGFVC method successfully fitted the half-Gaussian distributions and obtained stable thresholds for FVC estimation. The results indicate that the HAGFVC method can be used to effectively and accurately derive FVC images, with a small mean bias error (MBE) and with root mean square error (RMSE) of less than 0.04 in all cases. Comparatively, other methods we tested performed poorly (RMSE of up to 0.36) because of the abundance of mixed pixels in LARS images, especially at high altitudes above ground level (AGL) or in the case of moderate vegetation coverage. The results demonstrate the importance of developing image-processing methods that specially account for mixed pixels for LARS images. Simulations indicated that the theoretical accuracy (no errors in fitting the half-Gaussian distributions) of the HAGFVC method reflected an RMSE of less than 0.07. Additionally, this method provides a useful approach to efficiently estimating FVC by using LARS images over large areas.

Original languageEnglish
Pages (from-to)379-390
Number of pages12
JournalAgricultural and Forest Meteorology
Volume262
DOIs
Publication statusPublished - 15 Nov 2018

Fingerprint

vegetation cover
maize
crop
corn
crops
pixel
remote sensing
methodology
vehicle
method
unmanned aerial vehicles
vegetation
processing technology
histogram
vegetative growth
image processing
flight
developmental stages
image analysis
normal distribution

Cite this

Li, Linyuan ; Mu, Xihan ; Macfarlane, Craig ; Song, Wanjuan ; Chen, Jun ; Yan, Kai ; Yan, Guangjian. / A half-Gaussian fitting method for estimating fractional vegetation cover of corn crops using unmanned aerial vehicle images. In: Agricultural and Forest Meteorology. 2018 ; Vol. 262. pp. 379-390.
@article{ffd693fe23df469dbbfd79d1894536e3,
title = "A half-Gaussian fitting method for estimating fractional vegetation cover of corn crops using unmanned aerial vehicle images",
abstract = "Accurate estimates of fractional vegetation cover (FVC) using remotely sensed images collected using unmanned aerial vehicles (UAVs) offer considerable potential for field measurement. However, most existing methods, which were originally designed to extract FVC from ground-based remotely sensed images (acquired at a few meters above the ground level), cannot be directly used to process aerial images because of the presence of large quantities of mixed pixels. To alleviate the negative effects of mixed pixels, we proposed a new method for decomposing the Gaussian mixture model and estimating FVC, namely, the half-Gaussian fitting method for FVC estimation (HAGFVC). In this method, the histograms of pure vegetation pixels and pure background pixels are firstly fit using two half-Gaussian distributions in the Commission Internationale d'Eclairage (CIE) L*a*b* color space. Then, a threshold is determined based on the parameters of Gaussian distribution to generate a more accurate FVC estimate. We acquired low-altitude remote-sensing (LARS) images in three vegetative growth stages at different flight altitudes over a cornfield. The HAGFVC method successfully fitted the half-Gaussian distributions and obtained stable thresholds for FVC estimation. The results indicate that the HAGFVC method can be used to effectively and accurately derive FVC images, with a small mean bias error (MBE) and with root mean square error (RMSE) of less than 0.04 in all cases. Comparatively, other methods we tested performed poorly (RMSE of up to 0.36) because of the abundance of mixed pixels in LARS images, especially at high altitudes above ground level (AGL) or in the case of moderate vegetation coverage. The results demonstrate the importance of developing image-processing methods that specially account for mixed pixels for LARS images. Simulations indicated that the theoretical accuracy (no errors in fitting the half-Gaussian distributions) of the HAGFVC method reflected an RMSE of less than 0.07. Additionally, this method provides a useful approach to efficiently estimating FVC by using LARS images over large areas.",
keywords = "(UAV), Digital photography, Fractional vegetation cover (FVC), Half-Gaussian distribution, Histogram threshold, Low-altitude remote sensing (LARS), Unmanned aerial vehicle",
author = "Linyuan Li and Xihan Mu and Craig Macfarlane and Wanjuan Song and Jun Chen and Kai Yan and Guangjian Yan",
year = "2018",
month = "11",
day = "15",
doi = "10.1016/j.agrformet.2018.07.028",
language = "English",
volume = "262",
pages = "379--390",
journal = "Agricultural and Forest Meteorology",
issn = "0168-1923",
publisher = "Pergamon",

}

A half-Gaussian fitting method for estimating fractional vegetation cover of corn crops using unmanned aerial vehicle images. / Li, Linyuan; Mu, Xihan; Macfarlane, Craig; Song, Wanjuan; Chen, Jun; Yan, Kai; Yan, Guangjian.

In: Agricultural and Forest Meteorology, Vol. 262, 15.11.2018, p. 379-390.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A half-Gaussian fitting method for estimating fractional vegetation cover of corn crops using unmanned aerial vehicle images

AU - Li, Linyuan

AU - Mu, Xihan

AU - Macfarlane, Craig

AU - Song, Wanjuan

AU - Chen, Jun

AU - Yan, Kai

AU - Yan, Guangjian

PY - 2018/11/15

Y1 - 2018/11/15

N2 - Accurate estimates of fractional vegetation cover (FVC) using remotely sensed images collected using unmanned aerial vehicles (UAVs) offer considerable potential for field measurement. However, most existing methods, which were originally designed to extract FVC from ground-based remotely sensed images (acquired at a few meters above the ground level), cannot be directly used to process aerial images because of the presence of large quantities of mixed pixels. To alleviate the negative effects of mixed pixels, we proposed a new method for decomposing the Gaussian mixture model and estimating FVC, namely, the half-Gaussian fitting method for FVC estimation (HAGFVC). In this method, the histograms of pure vegetation pixels and pure background pixels are firstly fit using two half-Gaussian distributions in the Commission Internationale d'Eclairage (CIE) L*a*b* color space. Then, a threshold is determined based on the parameters of Gaussian distribution to generate a more accurate FVC estimate. We acquired low-altitude remote-sensing (LARS) images in three vegetative growth stages at different flight altitudes over a cornfield. The HAGFVC method successfully fitted the half-Gaussian distributions and obtained stable thresholds for FVC estimation. The results indicate that the HAGFVC method can be used to effectively and accurately derive FVC images, with a small mean bias error (MBE) and with root mean square error (RMSE) of less than 0.04 in all cases. Comparatively, other methods we tested performed poorly (RMSE of up to 0.36) because of the abundance of mixed pixels in LARS images, especially at high altitudes above ground level (AGL) or in the case of moderate vegetation coverage. The results demonstrate the importance of developing image-processing methods that specially account for mixed pixels for LARS images. Simulations indicated that the theoretical accuracy (no errors in fitting the half-Gaussian distributions) of the HAGFVC method reflected an RMSE of less than 0.07. Additionally, this method provides a useful approach to efficiently estimating FVC by using LARS images over large areas.

AB - Accurate estimates of fractional vegetation cover (FVC) using remotely sensed images collected using unmanned aerial vehicles (UAVs) offer considerable potential for field measurement. However, most existing methods, which were originally designed to extract FVC from ground-based remotely sensed images (acquired at a few meters above the ground level), cannot be directly used to process aerial images because of the presence of large quantities of mixed pixels. To alleviate the negative effects of mixed pixels, we proposed a new method for decomposing the Gaussian mixture model and estimating FVC, namely, the half-Gaussian fitting method for FVC estimation (HAGFVC). In this method, the histograms of pure vegetation pixels and pure background pixels are firstly fit using two half-Gaussian distributions in the Commission Internationale d'Eclairage (CIE) L*a*b* color space. Then, a threshold is determined based on the parameters of Gaussian distribution to generate a more accurate FVC estimate. We acquired low-altitude remote-sensing (LARS) images in three vegetative growth stages at different flight altitudes over a cornfield. The HAGFVC method successfully fitted the half-Gaussian distributions and obtained stable thresholds for FVC estimation. The results indicate that the HAGFVC method can be used to effectively and accurately derive FVC images, with a small mean bias error (MBE) and with root mean square error (RMSE) of less than 0.04 in all cases. Comparatively, other methods we tested performed poorly (RMSE of up to 0.36) because of the abundance of mixed pixels in LARS images, especially at high altitudes above ground level (AGL) or in the case of moderate vegetation coverage. The results demonstrate the importance of developing image-processing methods that specially account for mixed pixels for LARS images. Simulations indicated that the theoretical accuracy (no errors in fitting the half-Gaussian distributions) of the HAGFVC method reflected an RMSE of less than 0.07. Additionally, this method provides a useful approach to efficiently estimating FVC by using LARS images over large areas.

KW - (UAV)

KW - Digital photography

KW - Fractional vegetation cover (FVC)

KW - Half-Gaussian distribution

KW - Histogram threshold

KW - Low-altitude remote sensing (LARS)

KW - Unmanned aerial vehicle

UR - http://www.scopus.com/inward/record.url?scp=85050868092&partnerID=8YFLogxK

U2 - 10.1016/j.agrformet.2018.07.028

DO - 10.1016/j.agrformet.2018.07.028

M3 - Article

VL - 262

SP - 379

EP - 390

JO - Agricultural and Forest Meteorology

JF - Agricultural and Forest Meteorology

SN - 0168-1923

ER -