直方图
水下
衰减
人工智能
计算机视觉
像素
计算机科学
失真(音乐)
偏振滤光片
直方图匹配
直方图均衡化
对比度(视觉)
光学
地质学
物理
图像(数学)
光学滤波器
带宽(计算)
电信
海洋学
放大器
作者
Haofeng Hu,Pengfei Qi,Xiaobo Li,Zhenzhou Cheng,Tiegen Liu
标识
DOI:10.1088/1361-6463/abdc93
摘要
Abstract Underwater images always suffer from low contrast and inaccurate colors due to scattering and absorption by particles when the target light propagates through turbid water. In this paper, we first found that a lot of intensity space is occupied by fewer pixels, called ‘tails’, on both sides of the histograms for the red, green and blue channels of the image. Based on this histogram attenuation prior and taking account of the advantage of a polarization filter we proposed an effective polarimetric recovery method to enhance the underwater image quality, which includes a specially designed histogram processing method, named ‘cut-tail histogram stretching’. This processing overcomes the limitation of traditional histogram-based methods and can further improve the restoration performance. The experimental results corresponding to underwater scenes with different turbidities and colors show that the proposed method can simultaneously enhance the image contrast and reduce the color distortion to some extent, and thus realize clear underwater vision.
科研通智能强力驱动
Strongly Powered by AbleSci AI