A new image dehazing algorithm based on energy functional is proposed for the purpose of recovering scene radiance of haze degraded images. Considering the atmospheric degradation model, skylight is firstly estimated using two different methods whether the degraded images include the sky area or not, having better results than traditional ones. Then the white balance is conducted in order to simplify the scattering model and build new form of the airlight. And dark channel prior is performed to evaluate the dark channel map. The prior knowledge and the prior assumptions are used to construct the energy functional model between the dark channel map and the airlight, bringing norm and norm into the model for better computation. Meanwhile, the split Bregman method is employed to obtain the airlight of the image. Finally, the scene radiance is recovered by the model. Experiment results show that the algorithm is valid and robust for haze images and enhance the visibility with high efficiency.