Atmospheric pollution by PM2.5 is a serious problem now at Beijing, China and its neighboring countries. The de-hazing or de-fogging methods for degraded images have been a long-pending question at NASA Langley Research Center. Recently their basic Retinex model advanced into Visual Servo system. While the current main stream for unveiling the atmospheric pollution layer is based on scattering physics. Above all, a single image de-hazing model based on Dark Channel Prior hypothesis is most notable in practice. The keys to unveiling the pollution layer lie in the two points: [1] how to extract the skylight and [2] how to estimate the scene transmittance. This paper proposes a simple but effective de-hazing algorithm with banding-free and low computation costs referring to the Dark Channel Prior hypothesis. The simulation shows how the proposed model works to look the scene through heavy air pollution.