Browsing by Subject "Classification dataset"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Hyperspectral image dataset for salt stress phenotyping of wheat
    (2018-04-13) Moghimi, Ali; Yang, Ce; moghi005@umn.edu; Moghimi, Ali; Moghimi, Ali
    The dataset contains hyperspectral images of four wheat lines, each with a control and a salt (NaCl) treatment. Images were captured by a hyperspectral camera (PIKA II, Resonon) under natural light condition one day after salt application when there were no visual symptoms in wheat plants. The camera recorded the spectral response of both control and salt tanks of each line over 240 spectral channels in visible and near infrared region (400 nm to 900 nm) with about 2.1 nm spectral resolution, 640 spatial channels in the cross-track direction, and about 1 mm spatial resolution. Raw images were converted to radiance (Wm−2sr−1nm−1) using a vendor-provided calibration file, and then converted to reflectance (%) using a Spectralon panel. In total 25 spectral bands were disregarded due to high noise. Subsequent to noisy band removal, vegetation pixels were segmented from background using spectral vegetation indices and morphological operation. Although the goal of this study was plant phenotyping to rank salt tolerance of wheat lines, this dataset can be used for other research purposes, such as developing classification algorithms to discriminate healthy and stressed plants and developing methods for spectral feature selection to reduce the dimension of hyperspectral images.