摘要
传统去雾系统大多针对单幅图像进行处理,其稳定性和实时性不足,不仅算法复杂度较高,还存在亮光区域直方图过曝的缺陷,难以应用于实时视频图像修复中。针对该问题,设计了一种基于FPGA的1080P全高清视频图像实时增强去雾系统。首先,在RGB颜色空间采用分块分通道进行直方图统计;其次,通过预设阈值对多通道合并后的直方图进行削减,限制对比度过度放大,而后进行双线性插值计算;最后,对改进后的限制对比度自适应直方图均衡算法(contrast limited adaptive histogram equalization,CLAHE)进行仿真验证,并对系统进行有雾视频、图像以及户外场景的去雾测试。实验结果表明:本系统能实时处理帧速率为30 f/s、分辨率为1920×1080的图像,比传统方法去雾效果更好、更稳定,系统功耗低于5 W,可应用于低功耗领域。
Most defogging systems mainly focused on single image,lacking in stability and real-time performance.Besides,the traditional image haze removal algorithm had high complexity and the histogram of bright area was overexposed,it was difficult to be applied to real-time video image restoration.Facing this problems,a 1080P full-hd video image real-time enhanced defogging system based on FPGA was designed and constructed.At first,the histogram statistics was conducted in RGB color space by using blocks and dividing the channels.And then,the combined histogram of multi-channel was clips at a predetermined value,thereby limiting the contrast amplification.Finally,the bilinear interpolation is performed between the new histogram.We simulated the improved CLAHE algorithm,and tested the system for foggy video,image and outdoor scenes.Experimental results showed that the system can real-time process 30 frames per second 1920×1080 resolution images,compared with the mainstream methods,the defogging effect was better and more stable.The system power was less than 5 watts,so it could be applied to low power consumption.
作者
刘华军
张瑞珏
刘建锋
王盛
夏巧桥
LIU Huajun;ZHANG Ruijue;LIU Jianfeng;WANG Sheng;XIA Qiaoqiao(School of Physical Science and Technology,Central China Normal University,Wuhan 430079,China;Shanghai Institute of Space Electronics Technology,Shanghai 201109,China)
出处
《郑州大学学报（工学版）》
CAS
北大核心
2020年第2期19-24,共6页
Journal of Zhengzhou University（Engineering Science）
基金
国家自然科学基金资助项目(61501334)
中央高校基本科研业务费(CCNU16A05028)。
关键词
图像去雾
视频图像
FPGA
实时
低功耗
image dehazing
video image
FPGA
real-time
low-power dissipation
