搜索

x

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

基于照射_反射模型和有界运算的多谱段图像增强

毕国玲 续志军 赵建 孙强

引用本文:
Citation:

基于照射_反射模型和有界运算的多谱段图像增强

毕国玲, 续志军, 赵建, 孙强

Multispectral image enhancement based on irradiation-reflection model and bounded operation

Bi Guo-Ling, Xu Zhi-Jun, Zhao Jian, Sun Qiang
PDF
导出引用
  • 根据多尺度照射_反射模型, 结合广义有界运算模型和引导滤波, 能够有效地解决多谱段降质图像的增强问题. 算法采用自适应的引导滤波核函数作为环绕函数, 估计反映图像整体结构的不同尺度的低频照射分量; 利用有界广义对数比(general log-radio, GLR)模型加法代替Retinex理论中的对数变换运算; 再由GLR模型减法去除照射分量, 将不同尺度的反射分量从原始图像中分割出来; 对不同尺度反射分量的有效信息采用有界GLR模型乘法和加法进行融合, 有效地避免光晕伪影现象及越界现象的发生, 得到多尺度反射分量图像, 即最终的增强图像. 通过对可见光波段的低照度图像和雾霾图像、红外图像、X光医学图像四组多谱段降质图像实验分析, 以对比度和信息熵作为评价指标, 与同类算法进行了图像增强效果的定性和定量对比, 结果表明本文算法增强后的图像纹理和边缘细节更加丰富、对比度更高、视觉效果更佳, 可广泛地应用于多种图像增强领域.
    According to irradiation-reflection model, by combining the generalized bounded operation model with guide filter, the problem of enhancement for multispectral degraded images with blurred details can be effectively solved and the contrast and low signal-to-noise ratio can be improved. The multi-scale reflection component image, i.e., final enhanced image is obtained through the following procedures: using the adaptive different scales of guide filter function as surround function estimate reaction; separating out the high-low-frequency information; obtaining the different scales Irradiation images which react the overall structure of the image; using the bounded generalized logarithmic ratio (GLR) model addition to replace the Retinex logarithmic transformation; taking a similar logarithmic transformation to the original image to improve the contrast of the image and make the dark area of image details enhanced; again using GLR model subtraction to remove illuminate components from the original image to segment the different scales of the reflection image, thereby avoiding the loss of small details and the big details caused halo effect and noise interference. With four direction Sobel gradient image which reflects the comprehensive edge details of image information the adaptive gain function can be obtained. To avoid the smooth area noise amplification, by using the GLR model multiplication and addition to fuse the effective information of different scales images, the multi-scale reflection image, namely the final enhanced image are obtained. The effective suppression of the emergence of halo effect and computing overflow, which can retain a large number of image details; the comparision of subjective visual effect and the quantitative parameter analysis of the visible low illumination image, haze image, infrared image and X-ray medical images (a total of four groups of multispectral degraded images), the use of the contrast and entropy as evaluation indices, qualitative and quantitative comparison with a variety of image enhancement algorithms, show that the proposed algorithm strengthens and keeps the details of the image texture and edge, realizes the image contrast enhancement and the effective dynamic range compression, has a strong anti-noise ability, and can meet a variety of practical engineering image enhancement needs. The results of the study has been used in the infrared thermal imager, and good results have been achieved. The proposed algorithm is only for 8-bit grayscale image enhancement, and the color image enhancement will be studied in the future.
    • 基金项目: 国家自然科学基金(批准号: 60977001)资助的课题.
    • Funds: Project supported by the National Natural Science Foundation of China (Grant No. 60977001).
    [1]

    Liu S G, Chen J H, Fan H Y 2011 Chin. Phys. B 20 120305

    [2]

    Zhao W D, Zhao J, Xu Z J 2013 Acta Phys. Sin. 62 214204 (in Chinese) [赵文达, 赵建, 续志军 2013 物理学报 62 214204]

    [3]

    Tan R T 2008 IEEE Conference on Computer Vision and Pattern Recognition 2008 p1

    [4]

    He K, Sun J, Tang X 2009 IEEE Conference on Computer Vision and Pattern Recognition 2009 p1956

    [5]

    Yu D, Bao X D 2010 J. Biomed. Engineer. Res. 29 5 (in Chinese) [余岱, 鲍旭东 2010 生物医学工程研究 29 5]

    [6]

    Jia D Y, Ding T H 2005 Acta Phys. Sin. 54 4058 (in Chinese) [郏东耀, 丁天怀 2005 物理学报 54 4058]

    [7]

    Land E H 1977 Sci. Am. 237 108

    [8]

    Jobson D J, Rahman Z, Woodell G A 1997 IEEE Trans. Image Process. 6 451

    [9]

    Xu X, Chen Q, Wang P A, Sun H J, Xia D S 2008 J. Computer-Aided Design & Computer Graphics 20 1325 (in Chinese) [许欣, 陈强, 王平安, 孙怀江, 夏德深 2008 计算机辅助设计与图形学学报 20 1325]

    [10]

    Meylan L 2006 IEEE Trans. Image Process. 15 2820

    [11]

    Tang L, Zhao C X, Wang H N, Shao W Z 2008 J. Image and Graphics 13 264 (in Chinese) [唐磊, 赵春霞, 王鸿南, 邵文泽 2008 中国图象图形学报 13 264]

    [12]

    Fang S, Yang J R, Cao Y, Wu P F, Rao R Z 2012 J. Image and Graphics 17 748 (in Chinese) [方帅, 杨静荣, 曹洋, 武鹏飞, 饶瑞中 2012 中国图象图形学报 17 748]

    [13]

    Jourlin M, Phinoli J C 1989 J. Microscopy 156 33

    [14]

    Zhu R F, Jia H G, Wang C, Wei Q, Zhang T Y, Yu L Y 2014 Optics and Precision Engineering 22 1064 (in Chinese) [朱瑞飞, 贾宏光, 王超, 魏群, 张天翼, 虞林瑶 2014 光学精密工程 22 1064]

    [15]

    Wang R G, Zhu J, Yang W T, Fang S, Zhang X T 2010 Acta Electron. Sin. 38 1181 (in Chinese) [汪荣贵, 朱静, 杨万挺, 方帅, 张新彤 2010 电子学报 38 1181]

    [16]

    Deng G, Cahill L W, Tobin G R 1995 IEEE Trans. Image Process. 4 506

    [17]

    Deng G 2013 IEEE Trans. Image Process. 22 2903

    [18]

    Nielsen F, Nock R 2009 IEEE Trans. Inform. Theory 55 2882

    [19]

    Jia H G, Wu Z P, Zhu M C, Xuan M, Liu H 2013 Optics and Precision Engineering 21 3272 (in Chinese) [贾宏光, 吴泽鹏, 朱明超, 宣明, 刘慧 2013 光学精密工程 21 3272]

  • [1]

    Liu S G, Chen J H, Fan H Y 2011 Chin. Phys. B 20 120305

    [2]

    Zhao W D, Zhao J, Xu Z J 2013 Acta Phys. Sin. 62 214204 (in Chinese) [赵文达, 赵建, 续志军 2013 物理学报 62 214204]

    [3]

    Tan R T 2008 IEEE Conference on Computer Vision and Pattern Recognition 2008 p1

    [4]

    He K, Sun J, Tang X 2009 IEEE Conference on Computer Vision and Pattern Recognition 2009 p1956

    [5]

    Yu D, Bao X D 2010 J. Biomed. Engineer. Res. 29 5 (in Chinese) [余岱, 鲍旭东 2010 生物医学工程研究 29 5]

    [6]

    Jia D Y, Ding T H 2005 Acta Phys. Sin. 54 4058 (in Chinese) [郏东耀, 丁天怀 2005 物理学报 54 4058]

    [7]

    Land E H 1977 Sci. Am. 237 108

    [8]

    Jobson D J, Rahman Z, Woodell G A 1997 IEEE Trans. Image Process. 6 451

    [9]

    Xu X, Chen Q, Wang P A, Sun H J, Xia D S 2008 J. Computer-Aided Design & Computer Graphics 20 1325 (in Chinese) [许欣, 陈强, 王平安, 孙怀江, 夏德深 2008 计算机辅助设计与图形学学报 20 1325]

    [10]

    Meylan L 2006 IEEE Trans. Image Process. 15 2820

    [11]

    Tang L, Zhao C X, Wang H N, Shao W Z 2008 J. Image and Graphics 13 264 (in Chinese) [唐磊, 赵春霞, 王鸿南, 邵文泽 2008 中国图象图形学报 13 264]

    [12]

    Fang S, Yang J R, Cao Y, Wu P F, Rao R Z 2012 J. Image and Graphics 17 748 (in Chinese) [方帅, 杨静荣, 曹洋, 武鹏飞, 饶瑞中 2012 中国图象图形学报 17 748]

    [13]

    Jourlin M, Phinoli J C 1989 J. Microscopy 156 33

    [14]

    Zhu R F, Jia H G, Wang C, Wei Q, Zhang T Y, Yu L Y 2014 Optics and Precision Engineering 22 1064 (in Chinese) [朱瑞飞, 贾宏光, 王超, 魏群, 张天翼, 虞林瑶 2014 光学精密工程 22 1064]

    [15]

    Wang R G, Zhu J, Yang W T, Fang S, Zhang X T 2010 Acta Electron. Sin. 38 1181 (in Chinese) [汪荣贵, 朱静, 杨万挺, 方帅, 张新彤 2010 电子学报 38 1181]

    [16]

    Deng G, Cahill L W, Tobin G R 1995 IEEE Trans. Image Process. 4 506

    [17]

    Deng G 2013 IEEE Trans. Image Process. 22 2903

    [18]

    Nielsen F, Nock R 2009 IEEE Trans. Inform. Theory 55 2882

    [19]

    Jia H G, Wu Z P, Zhu M C, Xuan M, Liu H 2013 Optics and Precision Engineering 21 3272 (in Chinese) [贾宏光, 吴泽鹏, 朱明超, 宣明, 刘慧 2013 光学精密工程 21 3272]

  • [1] 张东凌, 卢姁, 张铭. 有界大洋对纬向风强迫的响应及共振. 物理学报, 2018, 67(8): 089201. doi: 10.7498/aps.67.20172225
    [2] 王殿伟, 韩鹏飞, 范九伦, 刘颖, 许志杰, 王晶. 基于光照-反射成像模型和形态学操作的多谱段图像增强算法. 物理学报, 2018, 67(21): 210701. doi: 10.7498/aps.67.20181288
    [3] 刘杰, 张建勋, 代煜. 基于多引导滤波的图像增强算法. 物理学报, 2018, 67(23): 238701. doi: 10.7498/aps.67.20181425
    [4] 王聪, 杨晶, 潘秀娟, 蔡高航, 赵巍, 张景园, 崔大复, 彭钦军, 许祖彦. 基于光参量放大相位共轭特性的图像修复与增强. 物理学报, 2017, 66(10): 104205. doi: 10.7498/aps.66.104205
    [5] 刘琦, 王丽丹, 段书凯. 一种基于忆阻交叉阵列的自适应三高斯模型及其在图像增强中的应用. 物理学报, 2017, 66(12): 127301. doi: 10.7498/aps.66.127301
    [6] 李鑫楠, 黄贺艳, 贾小宁, 马驷良. 基于指导滤波的图像盲复原算法. 物理学报, 2015, 64(13): 134202. doi: 10.7498/aps.64.134202
    [7] 陈丽娟, 鲁世平. 地球磁层电磁场中粒子引导中心漂移运动模型的周期轨. 物理学报, 2014, 63(19): 190202. doi: 10.7498/aps.63.190202
    [8] 文侨, 王凯歌, 邵永红, 屈军乐, 牛憨笨. 基于偏振滤波图像增强和动态散斑照明的宽场荧光层析显微镜. 物理学报, 2013, 62(3): 034203. doi: 10.7498/aps.62.034203
    [9] 李向正. Fisher方程的有界衰减振荡解. 物理学报, 2012, 61(17): 170507. doi: 10.7498/aps.61.170507
    [10] 冯俊, 徐伟, 顾仁财, 狄根虎. 有界噪声与谐和激励联合作用下Duffing-Rayleigh振子的Melnikov混沌. 物理学报, 2011, 60(9): 090507. doi: 10.7498/aps.60.090507
    [11] 陈帝伊, 申滔, 马孝义. 参数不定的旋转圆盘在有界扰动下混沌振动的滑模变结构控制. 物理学报, 2011, 60(5): 050505. doi: 10.7498/aps.60.050505
    [12] 朱希睿, 孟续军. 改进的含温有界原子模型对金的电子物态方程的计算. 物理学报, 2011, 60(9): 093103. doi: 10.7498/aps.60.093103
    [13] 邢真慈, 徐伟, 戎海武, 王宝燕. 有界噪声激励下带有时滞反馈的随机Mathieu-Duffing系统的响应. 物理学报, 2009, 58(2): 824-829. doi: 10.7498/aps.58.824
    [14] 牛玉俊, 徐 伟, 戎海武, 王 亮, 冯进钤. 非光滑周期扰动与有界噪声联合作用下受迫Duffing系统的混沌预测. 物理学报, 2008, 57(12): 7535-7540. doi: 10.7498/aps.57.7535
    [15] 梁炎斌, 郑亚建, 杨平保, 曹龙贵, 陆大全, 胡 巍, 郭 旗. 有界非局域非线性介质中空间光孤子传输的研究. 物理学报, 2008, 57(9): 5690-5698. doi: 10.7498/aps.57.5690
    [16] 雷佑铭, 徐 伟. 有界噪声和谐和激励联合作用下一类非线性系统的混沌研究. 物理学报, 2007, 56(9): 5103-5110. doi: 10.7498/aps.56.5103
    [17] 孙晓娟, 徐 伟, 马少娟. 含有界随机参数的双势阱Duffing-van der Pol系统的倍周期分岔. 物理学报, 2006, 55(2): 610-616. doi: 10.7498/aps.55.610
    [18] 戎海武, 王向东, 徐 伟, 方 同. 有界随机噪声激励下软弹簧Duffing振子的安全盆分叉. 物理学报, 2005, 54(10): 4610-4613. doi: 10.7498/aps.54.4610
    [19] 田明锋, 孟续军, 朱希睿, 姜旻昊, 王志刚. 含温有界原子模型下电子与离子碰撞激发和电离截面的理论研究. 物理学报, 2005, 54(10): 4673-4679. doi: 10.7498/aps.54.4673
    [20] 蔡朝洪, 徐振源, 须文波. 利用凹槽滤波引导混沌系统到周期解. 物理学报, 2001, 50(10): 1846-1850. doi: 10.7498/aps.50.1846
计量
  • 文章访问数:  2591
  • PDF下载量:  301
  • 被引次数: 0
出版历程
  • 收稿日期:  2014-11-17
  • 修回日期:  2014-12-24
  • 刊出日期:  2015-05-05

基于照射_反射模型和有界运算的多谱段图像增强

  • 1. 中国科学院长春光学精密机械与物理研究所, 长春 130033;
  • 2. 中国科学院大学, 北京 100049
    基金项目: 国家自然科学基金(批准号: 60977001)资助的课题.

摘要: 根据多尺度照射_反射模型, 结合广义有界运算模型和引导滤波, 能够有效地解决多谱段降质图像的增强问题. 算法采用自适应的引导滤波核函数作为环绕函数, 估计反映图像整体结构的不同尺度的低频照射分量; 利用有界广义对数比(general log-radio, GLR)模型加法代替Retinex理论中的对数变换运算; 再由GLR模型减法去除照射分量, 将不同尺度的反射分量从原始图像中分割出来; 对不同尺度反射分量的有效信息采用有界GLR模型乘法和加法进行融合, 有效地避免光晕伪影现象及越界现象的发生, 得到多尺度反射分量图像, 即最终的增强图像. 通过对可见光波段的低照度图像和雾霾图像、红外图像、X光医学图像四组多谱段降质图像实验分析, 以对比度和信息熵作为评价指标, 与同类算法进行了图像增强效果的定性和定量对比, 结果表明本文算法增强后的图像纹理和边缘细节更加丰富、对比度更高、视觉效果更佳, 可广泛地应用于多种图像增强领域.

English Abstract

参考文献 (19)

目录

    /

    返回文章
    返回