Aiming at the problems of color distor tion in bright region and manual parameter acquisition in the traditional channe l prior algorithm,an adaptive image dehazing algorithm based on improved dark ch annel prior is proposed.Firstly,the adaptive fast OSTU method is presented to se gment the bright and dark regions,and then the dark value of bright and dark reg ion is obtained by region.Secondly, according to the distribution of bright regi ons,the air light of different regions is estimated adaptively.Thirdly, through analyzing the histogram characteristics of haze images, the adaptive calculating method of gray concentration is put forward to calculate the dehazing coefficie nt.Fourthly,the color of output image is adjusted by adopting the method of colo r level adaptive adjustment.Finally,the advantages of proposed algorithm are ver ified by developing comparative experiments and analysis.The results show that:t he proposed algorithm does not need to set the dehazing parameters artificially and has better robustness,which is suitable for all kinds of haze image,and the dehazed image has natural color and high contrast.