基于自适应条件直方图均衡的红外图像细节增强算法  

Infrared Image Detail Enhancement Based on Adaptive Conditional Histogram Equalization

作　　者：马群 赵美蓉[1] 郑叶龙[1] 孙琳[2] 倪锋[2] MA Qun;ZHAO Meirong;ZHENG Yeong;SUN Lin;NI Feng(School of Precision Instrument and Opto-Electronics Engineering,Tianjin University,Tianjin 300072,China;Jinhang Institute of Technical Physics,Tianjin 300308,China)

机构地区：[1]天津大学精密仪器与光电子工程学院,天津300072 [2]天津津航技术物理研究所,天津300308

出　　处：《红外技术》2024年第1期52-60,共9页Infrared Technology

基　　金：教育部联合基金项目(8091B022117)。

摘　　要：红外图像普遍存在对比度低、细节不清晰、边缘特征不突出等问题。针对这些问题,本文提出了一种自适应条件直方图均衡的红外图像细节增强算法。采用引导滤波将红外图像分解为背景层和细节层;然后采用自适应阈值邻域条件直方图结合对比度受限直方图均衡方式,对背景层图像进行灰度压缩和对比度增强;接着利用引导滤波的中间计算结果构造滤噪掩模,在对细节层进行增强的同时有效滤除背景噪声;最后将背景层和细节层处理结果进行线性融合得到细节增强后红外图像。主观评价和客观数据计算表明,本文提出的红外图像细节增强算法无须手动调节参数即可实现对各类场景的自适应,可以在抑制噪声的前提下,有效增强图像细节,并提升图像整体对比度水平。对算法进行了嵌入式移植,显示效果和资源占用表明算法具有很强的工程化应用水平。There are many problems with infrared images,such as low contrast,unclear details,and non-prominent edge features.To solve these problems,this study proposes an adaptive conditional histogram equalization algorithm for infrared image detail enhancement.First,the infrared image is decomposed into background and detail layers by a guided filter.Second,the combined adaptive threshold neighborhood condition histogram and contrast limited histogram equalization method are used to compress and enhance the gray level of the background image.Then a noise mask is constructed using the intermediate calculation results of the guided filter,which can effectively filter the background noise while enhancing the detail layer.Finally,the background and detail layer processing results are linearly fused to obtain a detail-enhanced infrared image.Subjective evaluation and objective data calculation show that the infrared image detail enhancement algorithm proposed in this paper realizes adaptation to various scenes without manual parameter adjustment,and can effectively enhance the image details and improve the overall contrast level of the image under the premise of suppressing noise.Embedded transplantation of the algorithm was performed,and the display effect and resource occupation show that the algorithm has strong engineering application prospects.

关 键 词：红外图像增强 引导滤波 直方图均衡 嵌入式移植 

分 类 号：TN215[电子电信—物理电子学]