基于FPGA的车载图像去雾系统设计  

FPGA-based Vehicle Image Defogging System Design

作　　者：张强 杨志芳[1] 从桂浩 ZHANG Qiang;YANG Zhifang;CONG Guihao(School of Electrical and Information Engineering,Wuhan Institute of Technology,Wuhan 430223,China)

机构地区：[1]武汉工程大学电气信息学院,武汉430223

出　　处：《自动化与仪表》2025年第2期138-143,共6页Automation & Instrumentation

基　　金：武汉工程大学研究生教育创新基金项目(CX2023560)。

摘　　要：雾天环境下行车记录设备采集的车载图像存在细节丢失和天空区域颜色失真等问题,为了辅助行车道路安全,减少降质图像对行车分析的影响,需要对车载图像进行快速去雾处理。首先,针对传统暗通道去雾算法存在的天空区域色彩失真等问题,提出了一种基于梯度和局部香农熵融合的天空区域分割方法;其次,针对大气明亮区域的估计透射率不匹配及计算速率较慢等问题,提出了一种快速引导滤波方法细化实际透射率;最后,设计改进暗通道去雾算法的硬件电路,实现基于FPGA的车载图像去雾系统的硬件设计。测试结果表明,该系统能够有效还原车载图像原始色彩,提高图像细节丰富度,具有运算速度快、可移植性高等优点,可以满足行车记录设备雾天行驶的实时性需求。In the foggy environment,the vehicle-mounted images collected by the driving recording equipment have problems such as loss of details and color distortion in the sky area.In order to assist the driving road safety and reduce the impact of degraded images on driving analysis,it is necessary to quickly defogged the vehicle-mounted images.Firstly,aiming at the problem of color distortion in the sky region of the traditional dark channel dehazing algorithm,a sky region segmentation method based on gradient and local Shannon entropy fusion is proposed.Secondly,a fast guided filtering method is proposed to refine the actual transmittance in view of the mismatch of the estimated transmittance in the bright area of the atmosphere and the slow calculation rate.Finally,the hardware circuit of the improved dark channel defogging algorithm is designed to realize the hardware design of the vehicle image defogging system based on FPGA.The test results show that the system can effectively restore the original color of the vehicle image and improve the richness of image details.It has the advantages of fast operation speed and high portability,and can meet the real-time requirements of driving recording equipment in foggy days.

关 键 词：暗通道去雾 天空区域分割 快速引导滤波 FPGA 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]