Image Enhancement System Based on Improved Dark Channel Prior

Chang Liu, Jun Zhu, Xiaojun Peng
2015 Proceedings of the 2015 International Conference on Electromechanical Control Technology and Transportation   unpublished
In order to reduce the bad influence of hazes on the computer vision system, an image dehazing algorithm based on dark channel prior is proposed. Firstly, we analyze the dark channel prior based image dehazing algorithm. Since the complexity of the algorithm is great and the acquired images have halo artifacts effect, the neighborhood similarity method, which gets the difference of the value of dark channel between every pixel and its nearest eight pixels, is introduced. The pixel of minimal
more » ... pixel of minimal difference is redefined as new dark channel. Besides, the adaptive gray stretch and local contrast enhancement algorithms are proposed to increase brightness and definition of the video. Finally, realization of the improved image enhancement algorithms based on DM6467 is analyzed in detail. The results indicate that the method can effectively improve the image quality.
doi:10.2991/icectt-15.2015.51 fatcat:nvjmelgiinh4xcwrencccuu5bu