Aboveground biomass (AGB) and leaf area index (LAI) are very important in plant breeding and precision agriculture, as major indicators of crop growth status and grain yield. However, the traditional destructive sampling method is impractical and time-consuming for large breeding programs, especially for small breeding plots in the early generations as lacking plants. In this study, an unmanned aerial vehicle (UAV) based platform mounted with digital and multispectral cameras was used for acquiring digital and multispectral images over a wheat field at multiple time points before flowering. The AGB and LAI are measured through destructive harvesting after flights. Ortho-mosaics were built using the image sets with a commercial software and then segmented into individual plots. Several visible and spectral vegetation indexes were calculated from digital and multispectral ortho-mosaics for each plot, respectively. The AGB and LAI were estimated using visible, spectral and the combination of all indexes using partial least squares regression models (PLSRMs) and robust partial least squares regression models (RPLSRMs), respectively, following with the k-fold cross-validation. The results showed that the three groups of vegetation indexes could predict AGB and LAI quite well and had similar prediction performance at severalgrowth stages (R2 = 0.65 ~ 0.84, 0.62 ~ 0.86 and 0.64 ~ 0.89 for visual, spectral and combined indexes, respectively, for AGB; R2 = 0.69 ~ 0.94, 0.65 ~ 0.89, 0.69 ~ 0.91 for LAI). In general, RPLSRM slightly outperformed PLSRM for estimating AGB and LAI. In conclusion, this study had demonstrated that both visible and spectral index extracted from the UAV-based platform were reliable for estimating AGB and LAI, which can be applied in the high throughput phenotyping in breeding programs.