This work aims at imaging gamma radiation sources for security and industrial applications, such as radiation source searching. By using a collimator-less position-sensitive scintillation detector, an angular image of source intensity distribution can be estimated from measuring the photon distribution in the detector by means of image reconstruction. Two types of sparsity priors, the L2-norm of image and the entropy of image are introduced in a MAP reconstruction framework to further improve imaging performance. Monte Carlo simulations and experiments demonstrate the feasibility of the proposed design. MAP reconstruction effectively improves both image resolution and source position estimation accuracy, especially in low count cases. This design is attractive for combined merits of good image resolution, portability and high sensitivity comparing to collimator based gamma cameras when imaging limited number of point sources.