运动图像增强与网络环境下图像信息跨尺度分析与融合
|
摘要
本文对运动图像增强与网络环境下图像信息跨尺度分析与融合进行了研究。图像增强是指根据特定的需要突出图像中的重要信息,同时减弱以及去除不需要的信息。网络环境下图像信息融合将多个成像传感器同一时间或同一成像传感器不同时间对同一场景采集的多幅图像融合成一幅更精确、更可靠的图像。论文完成的主要工作如下：
(1)研究基于空域的运动图像增强。提出运动图像LOG算子锐化增强方法,采用LOG算子预提取图像的边缘信息,将边缘信息与原图像结合获取增强图像,实验结果表明采用LOG算子锐化增强后的图像在信息熵、平均梯度、边缘强度等客观评价指标上都有了一定的提高。
(2)研究基于频域的运动图像增强。提出了一种基于NSCT的自适应阈值去噪方法,阈值计算不仅与噪音方差相关还与系数的本地变化相关,针对不同系数采用不同的增强算子。实验结果表明该方法不仅能够有效地去除图像中的噪音,还能保存并有效地增强图像中的边缘信息,图像的PSNR值有了一定的提高。开发了一套图像去噪工具,实现了图像输入、图像多种方法去噪、图像客观质量评价以及图像输出展示等功能。
(3)研究了网络环境下的图像信息跨尺度分析与融合。研究了图像跨尺度融合的低频以及高频系数融合规则,提出了一种基于NSCT的图像信息融合方法。低频系数采用改进能量对比度的融合规则,不但考虑到了系数的能量与其周围能量的对比,还考虑了系数周围能量的分布；高频系数采用了Context的融合规则。实验结果表明采用该方法融合后的图像更清晰,且在平均交互熵、平均梯度、信息熵、互信息、QABF等客观评价指标上有一定程度的提高。
将改进的算法分别应用在了空间图像和标准库图像上进行了验证,证明了算法的可行性和有效性。
This thesis studies motion image enhancement and cross-scale analysis and fusion under network environment. Image enhancement highlights important information in the image based on the specific needs, while weakens and removes unwanted information. The image information fusion under network environment is to fuse images of the same scene obtained by multiple imaging sensors in one time or the same imaging sensors in different times into a more accurate, more reliable image. The main work is as follows:
(1) Study motion image enhancement based on spatial domain. Propose a LOG operator sharpening method for motion image, using LOG operator image for edge information extraction, combine the edge information with the original image to get the enhanced image, the experimental results show that the image sharpen by LOG operator get some extent improvement in objective evaluation index of entropy, average gradient and edge strength.
(2) Study motion image enhancement based on spatial domain. Propose a spatially adaptive threshold denoising and enhancing method based on NSCT, the threshold is not only related to the noise variance but also the local variance of coefficient, then uses different enhancement operator for different coefficients. The experimental results show that the method can not only remove the image noise and retain edge in the image information, but also retain and effectively enhance the edge details of the image, PSNR of image get some extent improvement. In this thesis develops a set of image fusion tools to achieve the image input function, image denoising by multiple method function, objective image quality evaluation function and image output display function.
(3)Study image information multi-scale analysis and fusion under network environment. Study the low frequency and high frequency fusion rule in image information multi-scale fusion, present an image information multi-scale fusion method based on NSCT.Low frequency coefficients use the improved energy contrast fusion rules, not only takes into account the coefficient of energy and its surrounding energy contrast, consideration is also given to the coefficients of the surrounding energy distribution; high frequency coefficients are using the context fusion rules. The experimental results show that the method can effectively capture the image edge information. The fused image after experiment is clearer and gets some extent improvement in objective evaluation index of cross entropy, average gradient, information entropy, mutual information, QABF.
The improved algorithm were applied in the motion image and standard library images and proved the feasibility and effectiveness of the algorithm
(1)研究基于空域的运动图像增强。提出运动图像LOG算子锐化增强方法,采用LOG算子预提取图像的边缘信息,将边缘信息与原图像结合获取增强图像,实验结果表明采用LOG算子锐化增强后的图像在信息熵、平均梯度、边缘强度等客观评价指标上都有了一定的提高。
(2)研究基于频域的运动图像增强。提出了一种基于NSCT的自适应阈值去噪方法,阈值计算不仅与噪音方差相关还与系数的本地变化相关,针对不同系数采用不同的增强算子。实验结果表明该方法不仅能够有效地去除图像中的噪音,还能保存并有效地增强图像中的边缘信息,图像的PSNR值有了一定的提高。开发了一套图像去噪工具,实现了图像输入、图像多种方法去噪、图像客观质量评价以及图像输出展示等功能。
(3)研究了网络环境下的图像信息跨尺度分析与融合。研究了图像跨尺度融合的低频以及高频系数融合规则,提出了一种基于NSCT的图像信息融合方法。低频系数采用改进能量对比度的融合规则,不但考虑到了系数的能量与其周围能量的对比,还考虑了系数周围能量的分布；高频系数采用了Context的融合规则。实验结果表明采用该方法融合后的图像更清晰,且在平均交互熵、平均梯度、信息熵、互信息、QABF等客观评价指标上有一定程度的提高。
将改进的算法分别应用在了空间图像和标准库图像上进行了验证,证明了算法的可行性和有效性。
This thesis studies motion image enhancement and cross-scale analysis and fusion under network environment. Image enhancement highlights important information in the image based on the specific needs, while weakens and removes unwanted information. The image information fusion under network environment is to fuse images of the same scene obtained by multiple imaging sensors in one time or the same imaging sensors in different times into a more accurate, more reliable image. The main work is as follows:
(1) Study motion image enhancement based on spatial domain. Propose a LOG operator sharpening method for motion image, using LOG operator image for edge information extraction, combine the edge information with the original image to get the enhanced image, the experimental results show that the image sharpen by LOG operator get some extent improvement in objective evaluation index of entropy, average gradient and edge strength.
(2) Study motion image enhancement based on spatial domain. Propose a spatially adaptive threshold denoising and enhancing method based on NSCT, the threshold is not only related to the noise variance but also the local variance of coefficient, then uses different enhancement operator for different coefficients. The experimental results show that the method can not only remove the image noise and retain edge in the image information, but also retain and effectively enhance the edge details of the image, PSNR of image get some extent improvement. In this thesis develops a set of image fusion tools to achieve the image input function, image denoising by multiple method function, objective image quality evaluation function and image output display function.
(3)Study image information multi-scale analysis and fusion under network environment. Study the low frequency and high frequency fusion rule in image information multi-scale fusion, present an image information multi-scale fusion method based on NSCT.Low frequency coefficients use the improved energy contrast fusion rules, not only takes into account the coefficient of energy and its surrounding energy contrast, consideration is also given to the coefficients of the surrounding energy distribution; high frequency coefficients are using the context fusion rules. The experimental results show that the method can effectively capture the image edge information. The fused image after experiment is clearer and gets some extent improvement in objective evaluation index of cross entropy, average gradient, information entropy, mutual information, QABF.
The improved algorithm were applied in the motion image and standard library images and proved the feasibility and effectiveness of the algorithm
引文
[1]洪日昌.多源图像信息融合算法及应用研究[D].中国科学技术大学,2008.
[2]陈一平.图像增强及其在视觉跟踪中的应用[D].国防科学技术大学,2011.
[3]朱立新,王平安,夏德深.基于梯度场均衡化的图像对比度增强[J].计算机辅助设计与图形学学报,2007,19(12)：1546-1552.
[4]熊杰.基于频率域、小波变换和神经网络的真彩图像增强算法研究[D].西北大学,2010.
[5]马志峰,吴琼之,杜娟.基于旋转、平移和尺度不变的平稳小波图像增强[J].光学技术,2009,35(01)：98-101.
[6]杨帆,赵瑞珍,胡绍海.基于Contourlet系数相关特性的自适应图像去噪算法[J].光学学报,2009,(02)：357-361.
[7]卓琳.基于Contourlet变换的图像融合算法[D].厦门大学,2008.
[8]李勇.基于多尺度分解的多源图像信息融合算法研究[D].吉林大学,2010.
[9]孙明超.可见光与红外侦察图像信息融合技术研究[D].中国科学院研究生院(长春光学精密机械与物理研究所),2012.
[10]袁文成；杨德兴；陈超.图像混合噪声的一种组合滤波消除方法[J].微处理机,2007,28(04)：78-83.
[11]梁伟,崔刚,陈桂琴.基于小波多尺度相关的特征边缘提取方法[J].电光与控制,2009,1 6(04)：86-89.
[12]刘燕.基于多尺度空间的局部方向去噪方法及其应用[D].山东大学,2008.
[13]卢萌,赵合计.一种改进的小波自适应边缘检测算法[J].计算机应用研究,201 1,28(01)：385-387.
[14]蒋年德.多尺度变换的图像融合方法与应用研究[D].湖南大学,2010.
[15]雷浩鹏,李峰.基于多小波-非采样Contourlet变换的自适应阈值图像去噪方法[J].计算机应用,2010,30(05)：1351-1355.
[16]曾业战,钱盛友,刘畅,王岐学,丁亚军.非下采样Contourlet变换自适应图像去噪方法[J].计算机工程与应用,2010,46(10)：157-159.
[17]屈小波,闫敬文,肖弘智,朱自谦.非降采样Contourlet域内空间频率激励的PCNN图像融合算法[J].自动化学报,2008,34(12)：1508-1514.
[18]宋文蕾,赵艳华,谢慕君.基于脊波变换的数字图像处理[J].长春大学学报,2007,17(01)：49-53.
[19]郑静,郑永果.一种基于Contourlet变换的图像融合方法[J].信息技术与信息 化,2011(02)：37-40.
[20]李栋,王敬东,李鹏.基于NSCT变换和小波变换相结合的图像融合算法研究[J].光电子技术,2011,31(02)：87-92.
[21]宋亚军,倪国强,高昆.基于小波-Contourlet变换的区域能量加权图像融合算法[J].北京理工大学学报,2008,28(02)：168-172.
[22]艾泽潭,石庚辰.小波变换在图像去噪中的应用[J].科技导报,2010,28(01)：101-106.
[23]杜超本,贾振红,覃锡忠,杨杰,胡英杰,李殿均.基于小波和NSCT的图像自适应阈值去噪方法[J].电讯技术,2011,51(05)：67-70.
[24]Cunha L D, Zhou Jian-ping, and Do M N. The nonsubsampled contourlet transform:theory, design, and applications[J]. IEEE Trans Image Processing,2006, 15(10):3089-3101.
[25]武海洋,王慧,程宝琴.基于最小Bayes风险的小波域局部自适应图像去噪[J].计算机应用,2010,30(12)：3238-3245.
[26]Eslami R and Radha H. "Translation-invariant Contourelt transform and its application to image denoising"[J]. IEEE Trans. Image Processing,2006,15(11): 3362-3374.
[27]Zhou Jian-ping, Cunha A L, and Do M N. Nonsubsampled contourlet transform: construction and application in enhancement[C]. IEEE Int. Conf. Image Processing, Genova, Italy, Sept.2005,1:11-14.
[28]S. G. Chang, B. Yu, and M. Vetterli, Spatially adaptive wavelet thresholding with context modeling for image denoising, IEEE Trans. Image Processing,2000,9(9): 1522-1531.
[29]S.G.Chang,B. Yu, and M. Vetterli.Adaptive Wavelet Thresholding for Image Denoising and Compression.IEEE Trans. Image Processing,2000,9(9):451-462.
[30]Chen, G. Y., Bui, T. D., Krzyzak.Image denoising using neighbouring wavelet coefficient.ICASSP'04. IEEE Int. Conf. Image Processing,2004,2:917-920
[31]S.G.Chang and M. Vetterli.Spatial adaptive wavelet thresholding for image denoising.IEEE Int. Conf. Image Processing,1997,2:374-377.
[32]Y. Yoo, A. Ortega, and B. Yu. Image subband coding using context based classification and adaptive quantization. IEEE Trans. Image Processing,1999,8:1702-1715.
[33]马常霞,赵春霞,胡勇,王鸿南,陈海燕.结合NSCT和图像形态学的路面裂缝检测[J].计算机辅助设计与图形学学报,2009,21(12)：1761-1767.
[34]王建华.基于Harr小波-Contourlet变换的图像增强算法[J].西北民族大学学报(自然科学版),2009,30(02)：50-54.
[35]张林,朱兆达.基于非降采样Contourlet变换的非线性图像增强新算法[J].电子与信息学报,2009,31(08)：1786-1790.
[36]向静波,苏秀琴,陆陶.基于Contourlet变换和形态学的图像增强方法[J].光子学报,2009,38(01)：224-227.
[37]苗启广,SE宝树.基于改进的拉普拉斯金字塔变换的图像融合方法[J].光学学报,2007,27(09)：1605-1610.
[38]叶传奇,苗启广,王宝树.基于非子采样Contourlet变换的图像融合方法[J].计算机辅助设计与图形学学报,2007,19(10)：1274-1278.
[39]叶传奇,王宝树,苗启广.基于NSCT变换的红外与可见光图像融合算法[J].系统工程与电子技术,2008,30(4)：593-596.
[40]杨谬,郭宝龙,倪伟.基于区域特性的Contourlet域多聚焦图像融合算法[J].西安交通大学学报,2007,41(4)：448-452.
[2]陈一平.图像增强及其在视觉跟踪中的应用[D].国防科学技术大学,2011.
[3]朱立新,王平安,夏德深.基于梯度场均衡化的图像对比度增强[J].计算机辅助设计与图形学学报,2007,19(12)：1546-1552.
[4]熊杰.基于频率域、小波变换和神经网络的真彩图像增强算法研究[D].西北大学,2010.
[5]马志峰,吴琼之,杜娟.基于旋转、平移和尺度不变的平稳小波图像增强[J].光学技术,2009,35(01)：98-101.
[6]杨帆,赵瑞珍,胡绍海.基于Contourlet系数相关特性的自适应图像去噪算法[J].光学学报,2009,(02)：357-361.
[7]卓琳.基于Contourlet变换的图像融合算法[D].厦门大学,2008.
[8]李勇.基于多尺度分解的多源图像信息融合算法研究[D].吉林大学,2010.
[9]孙明超.可见光与红外侦察图像信息融合技术研究[D].中国科学院研究生院(长春光学精密机械与物理研究所),2012.
[10]袁文成；杨德兴；陈超.图像混合噪声的一种组合滤波消除方法[J].微处理机,2007,28(04)：78-83.
[11]梁伟,崔刚,陈桂琴.基于小波多尺度相关的特征边缘提取方法[J].电光与控制,2009,1 6(04)：86-89.
[12]刘燕.基于多尺度空间的局部方向去噪方法及其应用[D].山东大学,2008.
[13]卢萌,赵合计.一种改进的小波自适应边缘检测算法[J].计算机应用研究,201 1,28(01)：385-387.
[14]蒋年德.多尺度变换的图像融合方法与应用研究[D].湖南大学,2010.
[15]雷浩鹏,李峰.基于多小波-非采样Contourlet变换的自适应阈值图像去噪方法[J].计算机应用,2010,30(05)：1351-1355.
[16]曾业战,钱盛友,刘畅,王岐学,丁亚军.非下采样Contourlet变换自适应图像去噪方法[J].计算机工程与应用,2010,46(10)：157-159.
[17]屈小波,闫敬文,肖弘智,朱自谦.非降采样Contourlet域内空间频率激励的PCNN图像融合算法[J].自动化学报,2008,34(12)：1508-1514.
[18]宋文蕾,赵艳华,谢慕君.基于脊波变换的数字图像处理[J].长春大学学报,2007,17(01)：49-53.
[19]郑静,郑永果.一种基于Contourlet变换的图像融合方法[J].信息技术与信息 化,2011(02)：37-40.
[20]李栋,王敬东,李鹏.基于NSCT变换和小波变换相结合的图像融合算法研究[J].光电子技术,2011,31(02)：87-92.
[21]宋亚军,倪国强,高昆.基于小波-Contourlet变换的区域能量加权图像融合算法[J].北京理工大学学报,2008,28(02)：168-172.
[22]艾泽潭,石庚辰.小波变换在图像去噪中的应用[J].科技导报,2010,28(01)：101-106.
[23]杜超本,贾振红,覃锡忠,杨杰,胡英杰,李殿均.基于小波和NSCT的图像自适应阈值去噪方法[J].电讯技术,2011,51(05)：67-70.
[24]Cunha L D, Zhou Jian-ping, and Do M N. The nonsubsampled contourlet transform:theory, design, and applications[J]. IEEE Trans Image Processing,2006, 15(10):3089-3101.
[25]武海洋,王慧,程宝琴.基于最小Bayes风险的小波域局部自适应图像去噪[J].计算机应用,2010,30(12)：3238-3245.
[26]Eslami R and Radha H. "Translation-invariant Contourelt transform and its application to image denoising"[J]. IEEE Trans. Image Processing,2006,15(11): 3362-3374.
[27]Zhou Jian-ping, Cunha A L, and Do M N. Nonsubsampled contourlet transform: construction and application in enhancement[C]. IEEE Int. Conf. Image Processing, Genova, Italy, Sept.2005,1:11-14.
[28]S. G. Chang, B. Yu, and M. Vetterli, Spatially adaptive wavelet thresholding with context modeling for image denoising, IEEE Trans. Image Processing,2000,9(9): 1522-1531.
[29]S.G.Chang,B. Yu, and M. Vetterli.Adaptive Wavelet Thresholding for Image Denoising and Compression.IEEE Trans. Image Processing,2000,9(9):451-462.
[30]Chen, G. Y., Bui, T. D., Krzyzak.Image denoising using neighbouring wavelet coefficient.ICASSP'04. IEEE Int. Conf. Image Processing,2004,2:917-920
[31]S.G.Chang and M. Vetterli.Spatial adaptive wavelet thresholding for image denoising.IEEE Int. Conf. Image Processing,1997,2:374-377.
[32]Y. Yoo, A. Ortega, and B. Yu. Image subband coding using context based classification and adaptive quantization. IEEE Trans. Image Processing,1999,8:1702-1715.
[33]马常霞,赵春霞,胡勇,王鸿南,陈海燕.结合NSCT和图像形态学的路面裂缝检测[J].计算机辅助设计与图形学学报,2009,21(12)：1761-1767.
[34]王建华.基于Harr小波-Contourlet变换的图像增强算法[J].西北民族大学学报(自然科学版),2009,30(02)：50-54.
[35]张林,朱兆达.基于非降采样Contourlet变换的非线性图像增强新算法[J].电子与信息学报,2009,31(08)：1786-1790.
[36]向静波,苏秀琴,陆陶.基于Contourlet变换和形态学的图像增强方法[J].光子学报,2009,38(01)：224-227.
[37]苗启广,SE宝树.基于改进的拉普拉斯金字塔变换的图像融合方法[J].光学学报,2007,27(09)：1605-1610.
[38]叶传奇,苗启广,王宝树.基于非子采样Contourlet变换的图像融合方法[J].计算机辅助设计与图形学学报,2007,19(10)：1274-1278.
[39]叶传奇,王宝树,苗启广.基于NSCT变换的红外与可见光图像融合算法[J].系统工程与电子技术,2008,30(4)：593-596.
[40]杨谬,郭宝龙,倪伟.基于区域特性的Contourlet域多聚焦图像融合算法[J].西安交通大学学报,2007,41(4)：448-452.