Logistic Regression-Based Spectral Band Selection for Tree Species Classification: Effects of Spatial Scale and Balance in Training Samples

Abstract

In this letter, we evaluated the pixel-level and plotlevel tree species classification of Scots Pine, Norway Spruce, and deciduous birch in a boreal forest using 64-band AisaEAGLE II hyperspectral data in a wavelength range of 400-1000 nm. First, band selection was performed using a sparse logistic regressionbased feature selection algorithm with pixel-level and plot-level data in case of balanced and imbalanced training data. This resulted in 8-11 selected hyperspectral bands, depending on the properties of the data used. We evaluated a tree species classification with 8-11 selected hyperspectral bands directly for a least squares support vector machine (LS-SVM)-based pixel-level classification with a relatively small training set size (0.5%-1.5% of the total data) and obtained an accuracy and kappa of around 93.50% and 0.90, respectively. These results are around 0.53%-0.94% points lower than those obtained using all of the hyperspectral bands. Second, one important wavelength region highlight by the selected bands was used to modify the sensor sensitivity configuration in the Leica Airborne Digital Sensor 40 (ADS40) multispectral sensor. Using a simulation model and the hyperspectral data, the modified and standard Leica ADS40 sensor responses were simulated and compared, and the modified system simulated response indicates a 3%-5% point improvement in the pixel-level and plot-level LS-SVM classification accuracy compared with the simulated responses of the standard Leica ADS40 band configuration.

Criteria Comparison for Classifying Peatland Vegetation Types Using In Situ Hyperspectral Measurements

“…This classifier has been successfully used with high dimensional data (gene selection in cancer classification [133], feature selection in remote sensing [28,29,134]). …”

“…The Regularized Logistic Regression (RLR) is the combination of a linear model (logistic regression) and a regularization term. It is usually used for feature selection (e.g., Pant, P. et al [28] applied it to reduce the 64 spectral bands from the hyperspectral AisaEAGLE II sensor to classify tree species in boreal forest using SVM; Pal, M. [29] applied it for reducing the 79 bands from the hyperspectral Digital Airborne Imaging Spectrometer (DAIS) sensor and the 220 bands from the hyperspectral Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) sensor to classify different land covers using SVM) is investigated in this paper as a classifier.…”

Thermal and hyperspectral imaging for Norway spruce (Picea abies) seeds screening

“…In more detail, we used the model presented in Cawley and Talbot (2006), including the Bayesian optimization of the regularization parameter. This feature selection approach has been used successfully for hyperspectral data classification in remote sensing context in Pal (2012) and Pant et al (2014).…”

Detection and Differentiation between Laurel Wilt Disease, Phytophthora Disease, and Salinity Damage Using a Hyperspectral Sensing Technique

“…The purpose of combining the early stage and late stage data to obtain wavebands could be used in the early and late stage at the same time because not all tree leaves show a lot of symptoms at the same time. Disease is transmitted depending on severity of disease and disease development stage [24,37]. Environmental factors and the population of the vectors also affect the dynamic disease transmission in plants [14,56,57].…”

“…Spectral data were selected in the visible and near infrared domain from 350 to 950 nm. Spectral data were averaged for both stages to 10 nm and 40 nm based on previous studies [36,37]The purpose of averaging data to 10 and 40 nm is to reduce the numerous wavelengths and also reduce noise and interference between wavebands. Additionally, waveband filters for 10 and 40 nm are available in the market and are reasonably priced for future sensor development.…”