Multi-energy computed tomography has attracted much attention in recent years. One of the key tasks focuses on developing efficient and accurate reconstruction algorithm of multi-channel image, especially for low-dose and incomplete dataset. This work proposed a tensor low rank and image sparsity based reconstruction method. The regularization functional is formed by introducing the weighted tensor nuclear norm and L0-norm on image gradient of each channel. The alternating direction method of multipliers is applied to develop the corresponding algorithm. Preliminary experiments on simulated dataset are performed, and both visual inspection and numerical criteria indicate the advantages of the proposed new method.