| With the development of human society, the environmental problems has become a hot focus in recent years. With the winter coming, the fog weather have emerged in most areas of China, which have caused a bad influence to our lives. Sometimes the fog is much too serious which makes the ordinary monitoring system lost the ability of working properly. The images capturing in this situation will occur serious image degradation. The edge details of the images are often blurred. Also, the color information of the images will deviate frequently. As a result, Image defogging technology have been a hot research focus in image processing field. It is of great practical significance to restore the images captured in fog weather.This thesis is mainly based on the atmospheric scattering model to set up the fog degradation. First of all, I study carefully He’s algorithm about single image defogging algorithm that has several limitations in the defogging process: when we find the dark channel of a image using the local window is fixed; when we get the atmospheric optical value is not accurate and it is very easily influenced by the non-sky area or the white noise with highlight; and when we use the soft matting to refine the transmission of a image taking a lot of time. So we proposed the the corresponding improvements. Then we establish a new degradation model of fog with the properties of polarized light. Finally, we have implemented the image rapidly defogging algorithm about three different angles of polarized image. That is the improvements about how to accurately estimate the value of atmospheric light and the transmission of image is fast optimized by the bilateral filter. And then we do the same experiment with the different image scenarios to prove the proposed algorithm that can improve the quality of a image of fog. Finally we proved that the algorithm has good robustness with doing some subjective and objective evaluation,making the comparison with other algorithms. |
PDF Full Text Request