基于暗元先验规律去雾后续处理研究

ON SUBSEQUENT TREATMENT OF DARK CHANNEL PRIOR-BASED IMAGE DEHAZING ALGORITHM

基于暗元先验规律去雾算法处理后得到的复原图像存在亮度暗、对比度低、局部细节信息不突出等问题。分析复原后图像的噪声放大程度与介质投射率的关系,在此基础上提出一种自适应剪切直方图均衡的方法,并将该方法应用到插值直方图均衡中,使得在噪声大的区域有小的增强幅度,在噪声小的区域有大的增强幅度,因此与传统的图像增强算法相比该算法能够得到明亮清晰的图像,并抑制了噪声。实验结果表明,该方法能够得到视觉效果更加优秀的处理结果。

The restored image obtained from image dehazing algorithm based on dark channel prior has problems of low brightness,low contrast and indistinct local details.A novel approach,named adaptive clip histogram equalisation,is proposed based on the analysis of the relationship between noise amplification extent and medium projecting rate of the image restored.The application of the approach in interpolation histogram equalisation makes the regions with more noises have lower enhancement rate,vice versa.So this approach can achieve a brighter and clearer image and restrain the noise compared with conventional image enhancement algorithms.Experimental results show that the image obtained by the approach has better visual effects.