基于引导滤波的多尺度自适应矿井低质图像增强方法  

Multi-scale adaptive low-quality mine image enhancementbased on guided filtering

作　　者：王媛彬[1] 李媛媛 齐景锋 吴华英 段誉 WANG Yuanbin;LI Yuanyuan;QI Jingfeng;WU Huaying;DUAN Yu(College of Electrical and Control Engineering,Xi’an University of Science and Technology,Xi’an 710054,China;Yulin Informatization Operation and Maintenace Branch of Shaanxi Shanmei Yubei Coal Industry Co.,Ltd.,Yulin 719000,China)

机构地区：[1]西安科技大学电气与控制工程学院,陕西西安710054 [2]陕西陕煤榆北煤业有限公司榆林信息化运维分公司,陕西榆林719000

出　　处：《西安科技大学学报》2022年第6期1214-1223,共10页Journal of Xi’an University of Science and Technology

基　　金：国家自然科学基金项目(52174198);陕西省科技厅项目(2019JQ-892)。

摘　　要：由于矿井中复杂的环境和光照条件,采集到的图像往往模糊不清,信息量不足。针对矿井低质图像中存在的问题,提出一种基于引导滤波的多尺度自适应图像增强方法。首先,利用具有良好保边特性的引导滤波将原始图像分解;根据对比度受限的自适应直方图均衡化来调整基础层的对比度,然后结合梯度域内改进的直方图均衡化和多尺度的细节增强调整细节层,减弱背景干扰;最后,鉴于图像线性重构时人工设置融合参数,处理效率低下且增强质量较差的问题,通过灰狼寻优算法进行自适应的参数选择,并引入加权思想,构造一种基于信息熵和峰值信噪比的适应度函数作为最终图像增强效果的评价指标。与经典的图像增强方法结果相比较,该方法将信息熵、峰值信噪比和结构相似度分别提高5.10%,5.43%和12.51%,在提高图像整体的对比度和清晰度的同时丰富细节信息,有效改善图像质量。Due to the complex environment and illumination conditions in the mine,the collected images are often fuzzy and lack of information.A multi-scale adaptive image enhancement algorithm based on guided filtering is proposed in this paper.The low-quality image is decomposed by guided filtering with good edge-preserving characteristics to obtain high frequency and low frequency components.The contrast of the low frequency layer is adjusted according to the contrast-limited adaptive histogram equalization.The high frequency part is processed based on the improved histogram and multi-scale detail enhancement strategy in the gradient domain to reduce the interference of the background.In the course of image reconstruction based on unsharp masking,for the low processing efficiency and poor quality of the image caused by manual parameter setting,the idea of weighting is introduced,and the fitness function which considered information entropy and peak signal-to-noise ratio are used as the fitness of the gray wolf algorithm(GWO)to optimize the fusion coefficient.Compared with the results of classical image enhancement methods,the information entropy,peak signal-to-noise ratio,and structural similarity are increased by 5.10%,5.43%,and 12.51%on average,respectively,which can effectively enhance the overall contrast and clarity of the image while enriching the detailed information to improve the image quality.

关 键 词：引导滤波 多尺度 自适应 反锐化掩模 灰狼优化 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]