Estimation of rice growth and nitrogen nutrition status using color digital camera image analysis

Non-destructive characterization of crop canopy is important to acquire timely and inexpensive information for crop management and yield prediction. The objective of the present study was to construct a non-destructive method for monitoring crop growth and nitrogen (N) nutrition status with digital camera image analysis. Digital images of rice canopies grown with four cultivars under various nitrogen treatments were captured by a digital camera periodically before heading stage and at the same time rice plants were sampled to measure LAI, shoot dry weight, and shoot N accumulation in 2006, 2007, and 2009. Canopy cover (CC) and ten color indices were calculated from digital camera images using image analysis program developed in Visual Basic version 6.0. More than eight color indices and CC showed significant correlations with growth and N nutrition parameters like LAI, shoot dry weight, and shoot N accumulation. CC revealed the highest correlations with all of them. CC expressed a curve-linear relationship with LAI, shoot dry weight, and shoot N accumulation (X‘s), being well fitted to a negative exponential function: CC = CCmax {1 − EXP (−k X)} with asymptote (CCmax) of unity. However, the nonlinear relationship of CC with LAI and shoot N accumulation except shoot dry weight was statistically different among rice cultivars. The statistically different relationships of CC with all the three parameters were found among N fertilization levels as k values increase with the increase of N fertilization level. To compensate the effects of rice cultivar and N level, stepwise multiple linear regression (SMLR) models including nonlinear relationship of CC and color indices were calibrated and validated. The selected SMLR models for LAI, shoot dry weight, and shoot N accumulation showed better performance than the models using only CC as a predictor variable. The selected SMLR models for LAI, shoot dry weight, and shoot N accumulation showed acceptable precision and its accuracy, indicating that conventional color digital camera could be employed for characterizing the growth and N nutrition status of rice non-destructively.

► Canopy cover (CC) and ten color indices were calculated from digital camera images.
► CC was negative-exponentially related to LAI, shoot dry weight (SDW), and shoot N content (Nup).
► CC relationships with LAI and Nup were different among rice cultivars and N fertilization levels.
► Model using CC and color indices as predictors was developed to compensate the cultivar and N effect.
► The developed models for LAI, SDW, and Nup showed acceptable precision and accuracy.