Abstract: As a complement to satellite observation, the technique of Unmanned Aerial Vehicle (UAV) shows great advantages in monitoring crop stages at a large scale because of its high spatiotemporal resolutions, low cost and risk in field measurements. With the objective to validate the availability of UAV in reality, this paper designed and established a low-altitude land surface UAV observation system consisted of a UAV platform and a multispectral camera to obtain images of winter wheat at its five growth stages. On basis of these images, this paper proposed a new approach that using time-series histograms of vegetation index to refine the threshold value in the VI-threshold method for extracting vegetated pixels from images. The results show that the UAV system is available to obtain surface images feasibly and the new vegetated pixel retrieval method can provide reasonable fractional vegetation cover (FVC) which is generally consistent with the five growth stages of the winter wheat. Moreover, this paper also investigated the spatial scaling effects of estimate FVC using the images from the UAV system.