Research On Fast Image Clarity Technology Based On Environmental Light Physical Model

The suspended particles in the sky can scatter and absorb the light to form fog and haze,which cause the contrast and color of the outdoor photography.In the application of outdoor images with higher quality requirements,such as the military situation monitoring,criminal crime details,the car in front of the obstacle observation,all need to do to lower image quality,reduction target details.It can significantly improve the image quality and make it better applied to the advanced image processing field.According to the condition of a single image haze degraded and the existing defogging method for the high time complexity to ambient light physical model as the foundation,foundation to fog in the dark channel,leads to rapid visual optimization to fog algorithm.The two schemes for atmospheric optical vector A,which is a combination of edge detection and LOG interval: Four Tree fast positioning of atmospheric optical vector A,and use skyline to more accurately determine the value of A;second is: first of all,on the threshold of single image segmentation to find the sky area,and combined with the atmospheric optical vector A two figure precise positioning of the circulation model.On the dark channel prior model to determine the initial transmittance using the least squares filtering based on improved optimization,can guarantee the complete details of the edge and is affected by the noise level,while suppressing the halo effect.And the expansion and corrosion are used to optimize the edge of projection graph.Then using the atmospheric light value A and the optimized transmittance,the application of the ambient light model is used to restore the clear and foggy image.Finally,the algorithm of this paper a large number of test images is tested to evaluate this algorithm from the objective and subjective two aspects,experimental results show that the improved algorithm to save the processing time of fog visual effects in the optimization of.