Compressive Sensing (CS) indicates new mechanism for hyperspectral imaging and practical hyperspectral compressive sensors have been designed to acquire fewer compressive measurements. However the numerical reconstruction of the hyperspectral data from the compressive measurements requires solving an ill-posed inverse problem and additional constraints are needed to seek a better solution. Based on the observation that in CS reconstruction quality can be improved from intelligent use of prior knowledge of the original data, we proposed an efficient new method to reconstruct Hyperspectral Images (HSI) in this paper. Our method, which exploit the HSI data structure characters of spatial 2D piecewise smoothness, low-rank property and adjacent spectrum correlation, have allowed to reconstruct HSI with compound regularizers. Moreover, an efficient numerical algorithm is developed for our method. The experimental results show that our method exhibits its superiority over other known state-of-the-art methods with higher reconstruction quality at the same measurement rates.