We demonstrate numerically the discrete image recovery via stochastic resonance in optically induced photonic lattices. The underlying signals are regularly reinforced at the expense of scattering noise with the interplay of the periodic potentials and the self-focusing nonlinearity. We founded that the energy redistribution tends to be periodic and the signal reinforcement is promoted with the help of periodic potentials. The lattice intensity levels, applied voltages, and correlation lengths are important parameters to influence the recovery effects. The dynamic nonlinear evolution including intensity and power spectrum is modeled according to the two-dimensional quasi-particle motion model. Our results suggest a potential technology to detect the noisy images.