摘要
针对经典红外图像压缩增强算法处理高温差(同一图像中景物温度差值大)红外图像时,多景物下对比度弱且细节难保持,以及针对不同场景图像需手动调整参数的问题,本文提出一种基于直方图峰度分割的红外图像压缩增强方法。该方法首先通过预处理改善红外原始图像灰度范围宽而有效灰度级少的问题;然后引入峰度概念以实现自适应直方图分割,从而保护各场景下细节;而后利用灰度级频率与梯度信息划分各场景对应子直方图的灰度范围,以提高图像整体对比度;最后利用由导向滤波得到的灰度梯度直方图分配子直方图内灰度级,以抑制噪声。仿真结果表明,与当前主流压缩增强算法相比,该方法能够有效处理高温差场景图像,具备较好的整体增强效果与各场景下细节保持能力,拥有良好的场景适应性与实时性。
In response to the issues of weak multi scenario contrast,difficulty in detail preservation,and manual parameter adjustment requirements across different scenes encountered by classical infrared image compression enhancement algorithms when processing high temperature difference(large intra image temperature variations)infrared images,a histogram kurtosis segmentation based infrared image compression enhancement method is proposed in this paper.Firstly,preprocessing is employed to alleviate the issues of wide gray level range and few effective gray levels in raw infrared images.Subsequently,the kurtosis concept is introduced to achieve adaptive histogram segmentation,thereby preserving details across various scenarios.The gray level frequency and gradient information are then utilized to divide gray ranges for sub histograms corresponding to different scenes,enhancing overall image contrast.Finally,gray gradient histograms obtained through guided filtering are employed to allocate gray levels within sub histograms for noise suppression.Simulation results demonstrate that compared with current mainstream compression enhancement algorithms,the proposed method effectively processes high temperature difference scenario images,exhibiting superior overall enhancement effects,and detail preservation capabilities across scenarios,along with excellent scene adaptability and real time performance.
作者
张熙来
王诗强
谢剑锋
张挺
张东方
杨德祎
ZHANG Xi-lai;WANG Shi-qiang;XIE Jian-feng;ZHANG Ting;ZHANG Dong-fang;YANG De-yi(Beijing Institute of Remote Sensing Equipment,Beijing 100854,China)
出处
《激光与红外》
北大核心
2025年第10期1633-1642,共10页
Laser & Infrared
关键词
红外图像
峰度分割
细节增强
动态范围压缩
infrared image
kurtosis segmentation
detail enhancement
dynamic range compression
