Leaf color can be used as an indicator of the plant healthiness, and thus digital image analysis may provide farmers and researchers with time- and resource-saving methods for diagnosing plant nutrient status. The digital images are dependent on the ambient light, therefore the color indices of digital images should be compensated for the difference of ambient light. The objectives of this study were to develope the calibration methods for color indices under variable irradiance condition. Four color panels were used for RGB (Red, Green, Blue) values and color indices calibration purpose. Reflectance for each panel was measured by spectro-radiometer with a waveband range of 300 - 1100 nm. The reflectance values of four color panels was used as a reference for calibration of RGB values and color indices. Using digital camera color images were taken for rice canopies together with reference panel that was set up at the level of canopy surface. Digital images were obtained form rice fields with variable nitrogen fertilization managements at active tillering to panicle initiation stage. The calibration coefficients for color image indices were calculated by using the linear regression equation between the pixel values of color image for reference panel and their known reflectance values. The determination coefficients (r2) of linear regression between non-calibrated mean B values of plant pixels in color image and shoot nitrogen contents in four rice varieties, Odae, Hwasung, Chucheong, and Ilpum, was 0.30, 0.27, 0.37, and 0.27, respectively, while the respective r2 values were increased to 0.79, 0.85, 0.77, and 0.53.by applying the calibration coefficient. These results imply that color digital image analysis could be a promising method for diagnosing nitrogen nutrition status of rice canopy.