摘要
针对Retinex算法去雾存在自适应能力差、容易产生光晕现象的缺点,通过对Retinex方法去雾后的图像依次做对比度拉伸、统计R/G/B分量像素亮度个数、用三次B样条拟合、计算曲线右侧的极值、确定天空区域分割阈值、得到分割图像,然后对分割后的非天空区域做色偏校正,对天空区域做均值化处理,最后将处理后的图像融合.实验结果表明,算法处理后峰值信噪比提高约9.14%,信息熵提升约22.2%,对比度提高约6.41%,说明改进后的算法效果较好.
In view of the shortcomings of Retinex algorithm in defogging,such as poor adaptability and easy to produce halo phenomenon,the contrast stretching of the image after defogging by Retinex method is carried out in turn,the number of pixel brightness of R/G/B component is counted,the cubic B-spline is fitted,the extreme value on the right side of the curve is calculated,the segmentation threshold of sky region is determined,and the segmentation images are obtained.Then,the color deviation of the non-sky region after segmentation is corrected,the sky region is averaged.Finally,the processed image is fused.The experimental results show that the PSNR is improved by about 9.14%,the entropy is improved by about 22.2%,and the contrast is improved by about 6.41%,which shows that the improved algorithm has a good effect.
作者
李竹林
李鹏翼
车雯雯
LI Zhulin;LI Pengyi;CHE Wenwen(College of Mathematics and Computer Science,Yan’an University,Yan’an 716000,Shaanxi China)
出处
《河南科学》
2023年第1期1-6,共6页
Henan Science
基金
国家自然科学基金(61761042)
国家级大创项目(202110719020)。
作者简介
李竹林(1972-),女,教授,硕士生导师,博士研究生,主要研究方向为数字图像处理;李鹏翼(2000-),男,硕士研究生,主要研究方向为人工智能;车雯雯(1995-),女,硕士研究生,主要研究方向为数字图像处理。
