基于内容特征的图像自适应压缩研究
|
摘要
图像压缩是数字图像传输和存储以及多媒体技术中的关键问题,自适应压缩算法是提高压缩率的一种有效方法。本文研究基于内容特征的图像自适应压缩方法,旨在根据图像不同区域的内容特征,采用不同的压缩方法,使图像在恢复效果和压缩率上都有一定的改善和提高。主要研究内容包括自适应分块,自适应位分配以及自适应分块后的块效应处理。
提出基于SUSAN算子的自适应分块方法。在自适应分块时,图像的内容特征用图像分块区域内的边缘信息量来表示,根据边缘信息量的多少将图像区域划分为平滑块和非平滑块两种类型。自适应分块的目的是对平滑块采用大尺寸分块压缩,对非平滑块采用小尺寸分块压缩,本论文最大分块尺寸为64~*64大小,最小分块尺寸为8~*8大小。自适应分块采用四叉树结构,四叉树分块的原则是基于SUSAN算子的边缘信息量提取。实验证明图像内容特征越平滑,基于SUSAN算子的自适应分块方法对图像压缩率提高的贡献就越大。
提出基于DCT变换域重要系数的自适应位分配方法。重要系数是指幅值大的DCT变换系数。对于平滑块采用对量化表修正的方法实现重要系数的自适应位分配,对非平滑块采用量化表排序方法实现对重要系数的自适应位分配,对自适应量化时产生的地址信息采用地址聚类的编码方法。实验证明对非平滑块采用地址聚类编码方法减少DCT变换域重要系数的地址信息量,整个图像压缩率相对于标准JPEG提高大约20%-25%。
基于SUSAN算子的自适应分块后图像会产生明显的块效应,采用基于人类视觉系统的块效应去除算法进一步提高图像的恢复质量。实验证明采用本论文的自适应压缩方法与标准JPEG压缩方法相比,在PSNR(峰值信噪比)的改善和压缩率的提高方面都有很好的表现。
Image compression is the key problem in digital image transmission and storage , Adaptive compression is a useful method to imcrease the compression ratio. The discussion of this thesis is mainly the adaptive compression based on the content characters of images whose purpose is to improve the compression performance and the decoded images performance. The algorithm is mainly composed of adaptive partition , adaptive quantization and deblocking algorithm.
The adaptive partition based on the SUSAN operator is divided into smooth block and non-smooth block according to the edge information. The edge information is used to measure the content characters. The biggest size of block is 64*64,and the smallest one is 8*8.The adaptive block adopts to quadtree segment .The method of edge detection is used to partition adaptively. Experiments have proved that the bigger the size of the smooth block is, the better the performance of the compression is.
Adaptive quantization is based on the important coefficients which is defined as the content characters of images for non-smooth block. Adaptive quantization for smooth block modifies the quantization step to reserve the important coefficients . Adaptive quantization for non-smooth block resort the standard quantization step to reserve the important coefficients. But these important coefficients are needed to be presented by address information. In this thesis we make the address information aggregate to be helpful to compress. By aggregating the address, the address information is cut down and the compression performance advances 20%-25% compared to standard JPEG algorithm.
As the result of adaptive partition, it exhibit noticeable blocking artifact near the block boundaries. The deblocking algorithm based on HVS is used to recover the missing image blocks. Experiments have proved the algorithm of this thesis is super to the standard JPEG algorithm in the improvement of PSNR and compress performance.
提出基于SUSAN算子的自适应分块方法。在自适应分块时,图像的内容特征用图像分块区域内的边缘信息量来表示,根据边缘信息量的多少将图像区域划分为平滑块和非平滑块两种类型。自适应分块的目的是对平滑块采用大尺寸分块压缩,对非平滑块采用小尺寸分块压缩,本论文最大分块尺寸为64~*64大小,最小分块尺寸为8~*8大小。自适应分块采用四叉树结构,四叉树分块的原则是基于SUSAN算子的边缘信息量提取。实验证明图像内容特征越平滑,基于SUSAN算子的自适应分块方法对图像压缩率提高的贡献就越大。
提出基于DCT变换域重要系数的自适应位分配方法。重要系数是指幅值大的DCT变换系数。对于平滑块采用对量化表修正的方法实现重要系数的自适应位分配,对非平滑块采用量化表排序方法实现对重要系数的自适应位分配,对自适应量化时产生的地址信息采用地址聚类的编码方法。实验证明对非平滑块采用地址聚类编码方法减少DCT变换域重要系数的地址信息量,整个图像压缩率相对于标准JPEG提高大约20%-25%。
基于SUSAN算子的自适应分块后图像会产生明显的块效应,采用基于人类视觉系统的块效应去除算法进一步提高图像的恢复质量。实验证明采用本论文的自适应压缩方法与标准JPEG压缩方法相比,在PSNR(峰值信噪比)的改善和压缩率的提高方面都有很好的表现。
Image compression is the key problem in digital image transmission and storage , Adaptive compression is a useful method to imcrease the compression ratio. The discussion of this thesis is mainly the adaptive compression based on the content characters of images whose purpose is to improve the compression performance and the decoded images performance. The algorithm is mainly composed of adaptive partition , adaptive quantization and deblocking algorithm.
The adaptive partition based on the SUSAN operator is divided into smooth block and non-smooth block according to the edge information. The edge information is used to measure the content characters. The biggest size of block is 64*64,and the smallest one is 8*8.The adaptive block adopts to quadtree segment .The method of edge detection is used to partition adaptively. Experiments have proved that the bigger the size of the smooth block is, the better the performance of the compression is.
Adaptive quantization is based on the important coefficients which is defined as the content characters of images for non-smooth block. Adaptive quantization for smooth block modifies the quantization step to reserve the important coefficients . Adaptive quantization for non-smooth block resort the standard quantization step to reserve the important coefficients. But these important coefficients are needed to be presented by address information. In this thesis we make the address information aggregate to be helpful to compress. By aggregating the address, the address information is cut down and the compression performance advances 20%-25% compared to standard JPEG algorithm.
As the result of adaptive partition, it exhibit noticeable blocking artifact near the block boundaries. The deblocking algorithm based on HVS is used to recover the missing image blocks. Experiments have proved the algorithm of this thesis is super to the standard JPEG algorithm in the improvement of PSNR and compress performance.
引文
[1]黄贤武,王加俊,李家华.数字图像处理与压缩编码技术.成都:电子科技大学出版社,2000.
[2]Rafael C.Gonzalez,Richard E.Woods(美)著.阮秋琦,阮宇智译.数字图像处理.北京:电子工业出版社,2003.
[3]郭武.基于小波的感兴趣区域编码理论及技术:(硕士学位论文).长沙:国防科学技术大学,2002.
[4]Wallace G K.The JPEG still picture compression standard.IEEE Trans,Consumer Electronic,1992,138(1):184-195.
[5]Leger A,Omachi T,Wallace G K.JPEG still picture compression algorithm.Optical Enginee ring,1991,30(?):94?-954.
[6]Wallace G K.The JPEG still picture compression standard.IEEE,Common,1991,34(4):30-44.
[7]张春田,苏育挺,张静.数字图像压缩编码.北京:清华大学出版社,2006.
[8]胡栋.静止图像编码的基本方法和国际标准.北京:北京邮电大学出版社,2003.
[9]张宁.JPEG2000中的感兴趣区域编码算法.计算机与数字工程,2003,31(5):24-27.
[10]Adams M D.The JPE62000 Still Image Compression Standard.ISO/IEC JTC 1/SC 29/W61N 2412,2001.
[11]张晓娣,刘贵忠,曾召华等.JPEG2000图像压缩编码系统及其关键技术.数字电视与数字视频,2001,54(8):13-17.
[12]Nasiopoulos P,Ward R,Morse D.Adaptive Compression Coding.IEEE Transactions on Communications,1991,39(8):1245-1254.
[13]Mahmoud R.EL- Sakka,Kamel M S.Adaptive Image Compression Based on Segmentation and Block Classification.IEEE,1998,2(4):555-558.
[14]then C.Adaptive Transform Coding Via Quadtree-Based Variable Blocksize DCT.IEEE International Conference,1989,13(5):1854-1857.
[15]Lee M,Crebbin G.Classified Vector Quantization with Variable Block-Size DCT models.IEE proceedings:Image and Signal Processing,1994,141(1):39-48.
[16]Vaisey J,Gersho A.Image Compression with Variable Block Size Segmentation.IEEE Transactions on Signal Processing,1992,40(8):2040-2060.
[17]Radha H,Vetterli M,Leonardi R.Image Compression Using Binary Space Partitioning Trees.IEEE Transactions on Image Processing,1996,5(12):1610-1624.
[18]Delp E J,Mitchell R.Image Compression Using Block Truncation Coding.IEEE Transactions,Commun,1979,27(6):1335-1342.
[19]Lema M D,Mitchell 0 R.Absolute moment block truncation coding and its application to color images.IEEE Trans,Commun,1984,32(12):1148-1157.
[20]魏海,沈兰荪.量化方法及其统计特征量用于图像检索的性能比较.电路与系统学报,2001,6(3):12-15.
[21]Ramamurthi B,Gersho A.Classified vector quantization of image.IEEE transactions on communications,1986,34(11):1105-1116.
[22]曹丽丹.基于SPIHT的无表零树图像压缩算法:(硕士学位论文).长春:吉林大学,2005.
[23]胡学龙,许开宇.数字图像处理.北京:电子工业出版社,2006.
[24]张旭东,卢国栋,冯健.图像编码基础和小波压缩技术——原理、算法和标准.北京:清华大学出版社,2004.
[25]Lee M H,Crebbin G.Image sequence coding using quadtree-based block-matching motion compensation and classified vector quantization.IEE Proc.-Vis.Image Signal Process,1994,141(6):453-460.
[26]Lee M H,Crebbin G.Classified vector quantization with variable block-size DCT models.IEE Proc.-Vis.Image Signal Process,1994,141(1):39-48.
[27]Chan M,Yu Y.Variable size block matching motion compensation with applications to video coding.Proc.Inst.Elect.Eng,1990,137(4):205-212.
[28]Chen W H,CHPratt.Scene Adaptive coder.IEEE Trans.Commun,1984,32(3):225-231.
[29]黄忠保.自适应菱形块图像编码分析.红外与激光工程,2000,29(3):32-34.
[30]欧阳林群.谈谈DCT在图像压缩编码中的应用原理.南平师范学报,2001,20(4):82-84.
[31]Chiptrasert B,Rao K R.Discrete cosine transform filtering.Signal Processing,1990,19(3):233-245.
[32]Andy Hung,Teresa H.-Y.Optimal Quantizer Step Size for Transform Coders.Standford University,1991,20(2):22-28.
[33]殷贤湘.基于四叉树的图像数据压缩.铁道师院学院(自然科学版),1995,12(4):16-24.
[34]马晓川,侯朝焕.新的纹理分类算法.中国图像学报,1999,4(5):387-390.
[35]Noureddine A,Djemel Z,Shengrui W.Autocovarian-base Perceptual Textural Features Corresponding to Human Visual Perception.Faculte des Sciences/ DMI Sherbrooke,Canada,1998,4(3):32-39.
[36]Noureddine A,Djemel Z,Takashi Y et al.Textural Features Corresponding to Visual Perception.IEEE Transactions,1978,8(6):460-473.
[37]姜玲,韩润萍,袁晔等.Canny 算子在织物疵点边缘检测中的应用.北京服装学院学报.2005,25(4):57-63.
[38]Canny J.A computational approach to edge detection.IEEE Trans Pattern Analysis and Machine Intelli2 gene,1986,8(6):679-698.
[39]方凯,方敏.基于Canny边缘检测的车牌字符分割方法.人工智能及识别技术,2007,32(9):1376-1380.
[40]Perez M M,Dennis Tim J.An Adaptive Implementation of the SUSAN Method for Image Edge and Feature Detection.Department of Electronic Systems Engineering,1997,2(26):394-397.
[41]马桂珍,房宗良,姚宗中.SUSAN边缘检测算法性能分析与比较.现代电子技术,2007,32(8):189-191.
[42]Golnerk M A,Mikhael W B.Region Based Variable Quantization for JPEG Image Compression.Krishnan School of Electrical Engineering and Computer Science University of Central Florida Orlando,2000,3(8):2816-2450.
[43]廖天康,苏彦锋,周荣政等.基于图像内容和人眼视觉特性的JPEG压缩编码.系统工程与电子技术,2002,24(2),93-102.
[44]唐跃林,肖沙里等.一种反映人眼视觉特性的JPEG压缩编码.微型机与应用,2004,(2):49-50.
[45]Siu-Wai W,Allen G.Rate-Constrained Picture-Adaptive Quantization For JPEG Baseline Coders.Department of Electrical and Computer Engineering University of California,1993,5(27):389-392.
[46]赵嵩.自适应数字图像压缩研究:(硕士学位论文).大连:大连海事大学,2007.
[47]黄战华,蔡怀宇,李贺桥等.自适应量化表的JPEG压缩技术.光电子.激光,2000,11(5):514-517.
[48]黄祥林,沈兰荪.基于DCT压缩域的纹理图像分类.电子与信息报,2002,24(2):216-221.
[49]http://sipi.usc.edu/database/
[50]谢胜利,石敏.基于DCT域的高压缩图像去块效应算法.控制理论与应 用,2006,23(1):139-142.
[51]徐金波.变换编码中消除图像”块效应”的优化算法.计算机工程与科学,2006,28(2):51-53.
[52]许志良.图像通信中质量提升技术的研究-图像后处理技术:(博士学位论文).广州:华南理工大学,2007.
[53]Reezne H,Lim J.Reduction of blocking effect in image coding,in Proc.ICASSP83,Boston,1983,8(14):1212-1215.
[54]Yang Y,Galatsanos N P,Katsaggelos A K.Regularized reconstruction to reduce blocking artifacts of block discrete cosine transform compressed images.IEEE Trans,1993,3(6):421-432.
[55]Stevenson R L.Reduction of coding artifacts in transform image coding.IEEE Int.Conf,Acoustics,Speech and Signal Processing,Minneapolis,1993,5(10):401-404.
[56]许志良,谢胜利.一种基于人类视觉系统的去块效应算法.电子与信息学报,2005,27(11):1715-1721.
[57]Zeng B.Reduction of blocking effect in DCT-coded images using zero-masking techniques.Signal Processing,1999,79(11):205-211.
[58]Lakhani G,Zhong N.Derivation of prediction equations for blocking effect reduction.IEEE Trans.on Circuits Sys.Video Technol,1999,9(3):415-418.
[59]Coudoux F X,Gzalet M,Corlay P.Reduction of blocking effect in DCT-coded images based on a visual perception criterion.Signal Processing:Image Commun,1998,11(3):179-186.
[60]Karunaseker S A,Kingsbury N G.A distortion measure for blocking artifacts in image based on human visual sensitivity.IEEE Trans.on Image Processing,1995,4(6):113-724.
[61]Lee J S.Digital image smoothing and the sigma filter.Computer Vision Graphics and Image Processing,1983,24(7):255-269.
[62]梁蔚华,基于自适应分块的视频图像压缩的研究:(硕士学位论文).大连,大连理工大学,2006.
[63]安丰玲,基于自适应分块的视频图像压缩研究:(硕士学位论文).大连,大连海事大学,2006。
[2]Rafael C.Gonzalez,Richard E.Woods(美)著.阮秋琦,阮宇智译.数字图像处理.北京:电子工业出版社,2003.
[3]郭武.基于小波的感兴趣区域编码理论及技术:(硕士学位论文).长沙:国防科学技术大学,2002.
[4]Wallace G K.The JPEG still picture compression standard.IEEE Trans,Consumer Electronic,1992,138(1):184-195.
[5]Leger A,Omachi T,Wallace G K.JPEG still picture compression algorithm.Optical Enginee ring,1991,30(?):94?-954.
[6]Wallace G K.The JPEG still picture compression standard.IEEE,Common,1991,34(4):30-44.
[7]张春田,苏育挺,张静.数字图像压缩编码.北京:清华大学出版社,2006.
[8]胡栋.静止图像编码的基本方法和国际标准.北京:北京邮电大学出版社,2003.
[9]张宁.JPEG2000中的感兴趣区域编码算法.计算机与数字工程,2003,31(5):24-27.
[10]Adams M D.The JPE62000 Still Image Compression Standard.ISO/IEC JTC 1/SC 29/W61N 2412,2001.
[11]张晓娣,刘贵忠,曾召华等.JPEG2000图像压缩编码系统及其关键技术.数字电视与数字视频,2001,54(8):13-17.
[12]Nasiopoulos P,Ward R,Morse D.Adaptive Compression Coding.IEEE Transactions on Communications,1991,39(8):1245-1254.
[13]Mahmoud R.EL- Sakka,Kamel M S.Adaptive Image Compression Based on Segmentation and Block Classification.IEEE,1998,2(4):555-558.
[14]then C.Adaptive Transform Coding Via Quadtree-Based Variable Blocksize DCT.IEEE International Conference,1989,13(5):1854-1857.
[15]Lee M,Crebbin G.Classified Vector Quantization with Variable Block-Size DCT models.IEE proceedings:Image and Signal Processing,1994,141(1):39-48.
[16]Vaisey J,Gersho A.Image Compression with Variable Block Size Segmentation.IEEE Transactions on Signal Processing,1992,40(8):2040-2060.
[17]Radha H,Vetterli M,Leonardi R.Image Compression Using Binary Space Partitioning Trees.IEEE Transactions on Image Processing,1996,5(12):1610-1624.
[18]Delp E J,Mitchell R.Image Compression Using Block Truncation Coding.IEEE Transactions,Commun,1979,27(6):1335-1342.
[19]Lema M D,Mitchell 0 R.Absolute moment block truncation coding and its application to color images.IEEE Trans,Commun,1984,32(12):1148-1157.
[20]魏海,沈兰荪.量化方法及其统计特征量用于图像检索的性能比较.电路与系统学报,2001,6(3):12-15.
[21]Ramamurthi B,Gersho A.Classified vector quantization of image.IEEE transactions on communications,1986,34(11):1105-1116.
[22]曹丽丹.基于SPIHT的无表零树图像压缩算法:(硕士学位论文).长春:吉林大学,2005.
[23]胡学龙,许开宇.数字图像处理.北京:电子工业出版社,2006.
[24]张旭东,卢国栋,冯健.图像编码基础和小波压缩技术——原理、算法和标准.北京:清华大学出版社,2004.
[25]Lee M H,Crebbin G.Image sequence coding using quadtree-based block-matching motion compensation and classified vector quantization.IEE Proc.-Vis.Image Signal Process,1994,141(6):453-460.
[26]Lee M H,Crebbin G.Classified vector quantization with variable block-size DCT models.IEE Proc.-Vis.Image Signal Process,1994,141(1):39-48.
[27]Chan M,Yu Y.Variable size block matching motion compensation with applications to video coding.Proc.Inst.Elect.Eng,1990,137(4):205-212.
[28]Chen W H,CHPratt.Scene Adaptive coder.IEEE Trans.Commun,1984,32(3):225-231.
[29]黄忠保.自适应菱形块图像编码分析.红外与激光工程,2000,29(3):32-34.
[30]欧阳林群.谈谈DCT在图像压缩编码中的应用原理.南平师范学报,2001,20(4):82-84.
[31]Chiptrasert B,Rao K R.Discrete cosine transform filtering.Signal Processing,1990,19(3):233-245.
[32]Andy Hung,Teresa H.-Y.Optimal Quantizer Step Size for Transform Coders.Standford University,1991,20(2):22-28.
[33]殷贤湘.基于四叉树的图像数据压缩.铁道师院学院(自然科学版),1995,12(4):16-24.
[34]马晓川,侯朝焕.新的纹理分类算法.中国图像学报,1999,4(5):387-390.
[35]Noureddine A,Djemel Z,Shengrui W.Autocovarian-base Perceptual Textural Features Corresponding to Human Visual Perception.Faculte des Sciences/ DMI Sherbrooke,Canada,1998,4(3):32-39.
[36]Noureddine A,Djemel Z,Takashi Y et al.Textural Features Corresponding to Visual Perception.IEEE Transactions,1978,8(6):460-473.
[37]姜玲,韩润萍,袁晔等.Canny 算子在织物疵点边缘检测中的应用.北京服装学院学报.2005,25(4):57-63.
[38]Canny J.A computational approach to edge detection.IEEE Trans Pattern Analysis and Machine Intelli2 gene,1986,8(6):679-698.
[39]方凯,方敏.基于Canny边缘检测的车牌字符分割方法.人工智能及识别技术,2007,32(9):1376-1380.
[40]Perez M M,Dennis Tim J.An Adaptive Implementation of the SUSAN Method for Image Edge and Feature Detection.Department of Electronic Systems Engineering,1997,2(26):394-397.
[41]马桂珍,房宗良,姚宗中.SUSAN边缘检测算法性能分析与比较.现代电子技术,2007,32(8):189-191.
[42]Golnerk M A,Mikhael W B.Region Based Variable Quantization for JPEG Image Compression.Krishnan School of Electrical Engineering and Computer Science University of Central Florida Orlando,2000,3(8):2816-2450.
[43]廖天康,苏彦锋,周荣政等.基于图像内容和人眼视觉特性的JPEG压缩编码.系统工程与电子技术,2002,24(2),93-102.
[44]唐跃林,肖沙里等.一种反映人眼视觉特性的JPEG压缩编码.微型机与应用,2004,(2):49-50.
[45]Siu-Wai W,Allen G.Rate-Constrained Picture-Adaptive Quantization For JPEG Baseline Coders.Department of Electrical and Computer Engineering University of California,1993,5(27):389-392.
[46]赵嵩.自适应数字图像压缩研究:(硕士学位论文).大连:大连海事大学,2007.
[47]黄战华,蔡怀宇,李贺桥等.自适应量化表的JPEG压缩技术.光电子.激光,2000,11(5):514-517.
[48]黄祥林,沈兰荪.基于DCT压缩域的纹理图像分类.电子与信息报,2002,24(2):216-221.
[49]http://sipi.usc.edu/database/
[50]谢胜利,石敏.基于DCT域的高压缩图像去块效应算法.控制理论与应 用,2006,23(1):139-142.
[51]徐金波.变换编码中消除图像”块效应”的优化算法.计算机工程与科学,2006,28(2):51-53.
[52]许志良.图像通信中质量提升技术的研究-图像后处理技术:(博士学位论文).广州:华南理工大学,2007.
[53]Reezne H,Lim J.Reduction of blocking effect in image coding,in Proc.ICASSP83,Boston,1983,8(14):1212-1215.
[54]Yang Y,Galatsanos N P,Katsaggelos A K.Regularized reconstruction to reduce blocking artifacts of block discrete cosine transform compressed images.IEEE Trans,1993,3(6):421-432.
[55]Stevenson R L.Reduction of coding artifacts in transform image coding.IEEE Int.Conf,Acoustics,Speech and Signal Processing,Minneapolis,1993,5(10):401-404.
[56]许志良,谢胜利.一种基于人类视觉系统的去块效应算法.电子与信息学报,2005,27(11):1715-1721.
[57]Zeng B.Reduction of blocking effect in DCT-coded images using zero-masking techniques.Signal Processing,1999,79(11):205-211.
[58]Lakhani G,Zhong N.Derivation of prediction equations for blocking effect reduction.IEEE Trans.on Circuits Sys.Video Technol,1999,9(3):415-418.
[59]Coudoux F X,Gzalet M,Corlay P.Reduction of blocking effect in DCT-coded images based on a visual perception criterion.Signal Processing:Image Commun,1998,11(3):179-186.
[60]Karunaseker S A,Kingsbury N G.A distortion measure for blocking artifacts in image based on human visual sensitivity.IEEE Trans.on Image Processing,1995,4(6):113-724.
[61]Lee J S.Digital image smoothing and the sigma filter.Computer Vision Graphics and Image Processing,1983,24(7):255-269.
[62]梁蔚华,基于自适应分块的视频图像压缩的研究:(硕士学位论文).大连,大连理工大学,2006.
[63]安丰玲,基于自适应分块的视频图像压缩研究:(硕士学位论文).大连,大连海事大学,2006。