Bioluminescence tomography (BLT) aims to reconstruct the internal bioluminescent source distribution with known optical properties and the measurements of light intensity on the surface of biological tissues. Due to the complexity of near-infrared photons transportation in biological tissues and the limited boundary measurements, the source reconstruction of BLT is a challenging problem. To obtain stable reconstruction with a small amount of measurements, we present a single-view based reconstruction method for multispectral BLT. In this method, we combine an iteratively shrinking permissible region strategy with sparsity-inducing regularization technique to deal with the high ill-posedness of BLT inverse problem. Simulations based on homogeneous numerical phantom are designed to systematically assess the performance of the proposed method in terms of localization accuracy and the ability to resolve two neighboring targets. We investigate single spherical sources with 3 sizes at 4 different depths, and double sources with 3 source separation distances varying from 3 to 9 mm. Simulation results show that based on single view measurement the proposed reconstruction method can yield stable reconstruction with an average 0.5 mm accuracy in single source cases and resolve two neighboring sources with 3 mm separation distance.