Abstract: Aiming at the problems that merging of low probability gray levels and loss of total gray levels lead to losing details and gray scale resolution, an improved HE model is proposed which takes the concept of removal of gray level not occupied in the image and limitation in the non-zero gray level, which guarantees the grayscale doesn’t decrease, and solves problems on gray-level preserving. Experiment shows that the model can obtain a good result for degraded images in mist.

Key words: histogram equalization, image enhancement, mist condition, gray-level preserving

摘要： 传统灰度图像直方图均衡模型在处理薄雾条件下的降质图像时，会出现概率较低的灰度级合并，总的灰度级有所损失问题，因此导致图像细节丢失，灰度分辨率下降，针对这一问题提出了改进的灰度图像直方图均衡模型，采取去除原始图像实际没有占有的灰度级，只对不为0的灰度级进行直方图均衡，从而保证总的灰度级数不减少，可以在一定程度上解决灰度级保持的问题。实验表明该直方图均衡模型可以对薄雾条件下降质图像处理取得较好的增强效果。

关键词: 直方图均衡, 图像增强, 薄雾条件, 保持灰度级