Research on Infrared Image Enhancement Method Combined with Single-scale Retinex and Guided Image Filter
-
摘要: 针对传统红外图像增强算法中图像对比度低、细节信息丢失与过度增强等问题,提出了一种单尺度Retinex与引导滤波相联合的红外图像增强方法。首先根据Retinex算法,利用主特征提取法获取原始图像的照射分量和反射分量,对照射分量采用平台直方图增强其对比度;然后利用局部方差加权引导滤波将反射分量分解为基本层和细节层,对两层分量的图像分别进行对比度和细节增强操作;最后将各个层次的结果按照合适的权重因子进行融合得到增强红外图像。实验结果表明,相比于其他增强算法,本文所提方法能更有效地提高红外图像的整体对比度,突出其细节特征,增强后的3组图像的信息熵和平均梯度平均值分别为9.7373和5.6922,相较于原图像分别提升了2.7499和3.8296。
-
关键词:
- 红外图像 /
- 图像增强 /
- 单尺度Retinex /
- 引导滤波 /
- 主特征提取
Abstract: This study proposes an infrared image enhancement method combined with single scale Retinex and guided image filtering to eliminate the problems of low image contrast, loss of detail information, and excessive enhancement in traditional infrared image enhancement algorithms. First, a structure extraction algorithm is used to obtain the incident and reflected components of the original image according to the Retinex algorithm. The platform histogram is used to enhance the contrast of the incident component. Then, the reflected component is decomposed into the base layer and detail layer by the weighted guided image filter based on variance and by performing contrast and detail enhancement operations on the images of the two components, respectively. Finally, the results of each level are fused according to the appropriate weight factors to obtain an enhanced infrared image. The experiments in this study show that the proposed method can improve the overall contrast of infrared images and highlight their detailed features more effectively than other enhancement algorithms. The information entropy and average gradient of the three groups of images after enhancement are 9.7373 and 5.6922, respectively, which are 2.7499 and 3.8296 higher than the original image. -
表 1 不同方法增强结果的信息熵
Table 1. Information entropy of results enhanced by different methods
Image Original image HE CLAHE BF Proposed method First scene 6.7363 7.9697 7.6739 7.0151 9.2159 Second scene 6.9593 7.9340 7.3236 7.0617 10.1542 Third scene 7.2665 7.9688 7.4556 7.3433 9.8417 表 2 不同方法增强结果的平均梯度
Table 2. Average gradient of results enhanced by different methods
Image Original image HE CLAHE BF Proposed method First scene 2.9937 6.9885 5.2015 3.8958 7.0755 Second scene 1.2590 2.2890 2.2865 1.5012 5.1617 Third scene 1.3350 2.0627 2.3768 1.5565 4.8394 -
[1] 汪子君, 罗渊贻, 蒋尚志, 等. 基于引导滤波的自适应红外图像增强改进算法[J]. 光谱学与光谱分析, 2020, 40(11): 3463-3467. https://www.cnki.com.cn/Article/CJFDTOTAL-GUAN202011026.htmWANG Zijun, LUO Yuanyi, JIANG Shangzhi, et al. An improved algorithm for adaptive infrared image enhancement based on guided filtering[J]. Spectroscopy and Spectral Analysis, 2020, 40(11): 3463-3467. https://www.cnki.com.cn/Article/CJFDTOTAL-GUAN202011026.htm [2] 路皓翔, 刘振丙, 郭棚跃, 等. 多尺度卷积结合自适应双区间均衡化的图像增强[J]. 光子学报, 2020, 49(10): 158-172. https://www.cnki.com.cn/Article/CJFDTOTAL-GZXB202010019.htmLU Haoxiang, LIU Zhenbing, GUO Pengyue, et al. Multi-scale convolution combined with adaptive bi-interval equalization for image enhancement[J]. Acta Photonica Sinica, 2020, 49(10): 158-172. https://www.cnki.com.cn/Article/CJFDTOTAL-GZXB202010019.htm [3] 葛朋, 杨波, 韩庆林, 等. 一种基于引导滤波图像分层的红外图像细节增强算法[J]. 红外技术, 2018, 40(12): 1161-1169. http://hwjs.nvir.cn/article/id/hwjs201812008GE Peng, YANG Bo, HAN Qinglin, et al. Infrared image detail enhancement algorithm based on hierarchical processing by guided image filter[J]. Infrared Technology, 2018, 40(12): 1161-1169. http://hwjs.nvir.cn/article/id/hwjs201812008 [4] 丁畅, 董丽丽, 许文海. "直方图"均衡化图像增强技术研究综述[J]. 计算机工程与应用, 2017, 53(23): 12-17. doi: 10.3778/j.issn.1002-8331.1710-0031DING Chang, DONG Lili, XU Wenhai. Review of "histogram" equalization technique for image enhancement[J]. Computer Engineering and Applications, 2017, 53(23): 12-17. doi: 10.3778/j.issn.1002-8331.1710-0031 [5] 徐超, 何利民, 王霞, 等. 红外偏振成像系统高速处理模块设计[J]. 红外与激光工程, 2017, 46(2): 133-140. https://www.cnki.com.cn/Article/CJFDTOTAL-HWYJ201702021.htmXU Chao, HE Limin, WANG Xia, et al. Design of high speed processing module for infrared polarization imaging system[J]. Infrared and Laser Engineering, 2017, 46(2): 133-140. https://www.cnki.com.cn/Article/CJFDTOTAL-HWYJ201702021.htm [6] 杨卫中, 徐银丽, 乔曦, 等. 基于对比度受限直方图均衡化的水下海参图像增强方法[J]. 农业工程学报, 2016, 32(6): 197-203. https://www.cnki.com.cn/Article/CJFDTOTAL-NYGU201606027.htmYANG Weizhong, XU Yinli, QIAO Xi, et al. Method for image intensification of underwater sea cucumber based on contrast limited adaptive histogram equalization[J]. Transactions of the Chinese Society of Agricultural Engineering, 2016, 32(6): 197-203. https://www.cnki.com.cn/Article/CJFDTOTAL-NYGU201606027.htm [7] 李佳, 李少娟, 段小虎, 等. 基于Retinex理论与概率非局部均值的红外图像增强方法[J]. 光子学报, 2020, 49(4): 187-196. https://www.cnki.com.cn/Article/CJFDTOTAL-GZXB202004021.htmLI Jia, LI Shaojuan, DUAN Xiaohu, et al. Infrared image enhancement based on Retinex and probability nonlocal means filtering[J]. Acta Photonica Sinica, 2020, 49(4): 187-196. https://www.cnki.com.cn/Article/CJFDTOTAL-GZXB202004021.htm [8] 王卫星, 赵恒. 结合改进Retinex及自适应分数阶微分的雾霾公路交通图像增强[J]. 光学精密工程, 2020, 28(8): 1820-1834. https://www.cnki.com.cn/Article/CJFDTOTAL-GXJM202008021.htmWANG Weixing, ZHAO Heng. Haze traffic image enhancement based on improved retinex and adaptive fractional differential[J]. Optics and Precision Engineering, 2020, 28(8): 1820-1834. https://www.cnki.com.cn/Article/CJFDTOTAL-GXJM202008021.htm [9] 王晓柱, 钮赛赛, 张凯, 等. 基于小波变换与特征提取的红外弱小目标图像融合[J]. 西北工业大学学报, 2020, 38(4): 723-732. doi: 10.3969/j.issn.1000-2758.2020.04.005WANG Xiaozhu, NIU Saisai, ZHANG Kai, et al. Image fusion of infrared weak-small target based on wavelet transform and feature extraction[J]. Journal of Northwestern Polytechnical University, 2020, 38(4): 723-732. doi: 10.3969/j.issn.1000-2758.2020.04.005 [10] Branchitta F, Diani M, Corsini G, et al. New technique for the visualization of high dynamic range infrared images[J]. Optical Engineering, 2009, 48(9): 096401-1-9. doi: 10.1117/1.3216575 [11] 葛朋, 杨波, 洪闻青, 等. 一种结合PE的高动态范围红外图像压缩及细节增强算法[J]. 红外技术, 2020, 42(3): 279-285. http://hwjs.nvir.cn/article/id/hwjs202003011GE Peng, YANG Bo, HONG Wenqing, et al. Dynamic range compression and detail enhancement algorithm combined with PE for high dynamic range infrared images[J]. Infrared Technology, 2020, 42(3): 279-285. http://hwjs.nvir.cn/article/id/hwjs202003011 [12] 周志强, 汪渤, 李立广, 等. 基于双边与高斯滤波混合分解的图像融合方法[J]. 系统工程与电子技术, 2016, 38(1): 8-13. https://www.cnki.com.cn/Article/CJFDTOTAL-XTYD201601003.htmZHOU Zhiqiang, WANG Bo, LI Liguang, et al. Image fusion based on a hybrid decomposition via bilateral and Gaussian filters[J]. Systems Engineering and Electronics, 2016, 38(1): 8-13. https://www.cnki.com.cn/Article/CJFDTOTAL-XTYD201601003.htm [13] HE Kaiming, SUN Jian, TANG Xiaoou. Guided image filtering[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35(6): 1397-1409. doi: 10.1109/TPAMI.2012.213 [14] LI Zhengguo, ZHENG Jinghong, ZHU Zijian, et al. Weighted guided image filtering[J]. IEEE Transactions on Image Processing, 2015, 24(1): 120-129. doi: 10.1109/TIP.2014.2371234 [15] 张庆宇, 范玉刚, 高阳. 基于单尺度Retinex与改进的K-均值聚类的涡流热成像缺陷检测[J]. 红外技术, 2020, 42(10): 1001-1006. http://hwjs.nvir.cn/article/id/hwjs202010014ZHANG Qingyu, FAN Yugang, GAO Yang. Defect detection of eddy-current thermography based on single-scale Retinex and improved K-means clustering[J]. Infrared Technology, 2020, 42(10): 1001-1006. http://hwjs.nvir.cn/article/id/hwjs202010014 [16] XU Li, YAN Qiong, XIA Yang, et al. Structure extraction from texture via relative total variation[J]. ACM Transactions on Graphics (TOG), 2012, 31(6): 1-10. [17] 李红, 吴炜, 杨晓敏, 等. 基于主特征提取的Retinex多谱段图像增强[J]. 物理学报, 2016, 65(16): 61-76. https://www.cnki.com.cn/Article/CJFDTOTAL-WLXB201616008.htmLI Hong, WU Wei, YANG Xiaomin, et al. Multispect ral image enhancement based on Retinex by using structure extraction[J]. Acta Phys, 2016, 65(16): 1-16. https://www.cnki.com.cn/Article/CJFDTOTAL-WLXB201616008.htm