An AOTF (Acousto-Optic Tunable Filter)-based spectral imager for field use was developed for hyper-spectral measurement of plant reflectance. Hyper-spectral imagery in a region between 450nm-900nm was obtained at 3 to 5nm resolution. Spectral wavelengths and interval were both programmable by PC control. Regression trial for rice canopies suggested that the leaf nitrogen and chlorophyll contents would be estimated by high-resolution measurement in visible and near infrared region. The directional multi-spectral measurements in visible, near-, and shortwave- infrared regions were obtained for rice canopies and used for inversion of bi-directional reflectance models. Results suggested that physically based directional model would be used for directional correction and for estimation of eco-physiological variables of rice canopies although they seemed less accurate in red region under sparse condition.