CN102148993B - Method and device for encoding wavelet image - Google Patents

Method and device for encoding wavelet image Download PDF

Info

Publication number
CN102148993B
CN102148993B CN 201010116325 CN201010116325A CN102148993B CN 102148993 B CN102148993 B CN 102148993B CN 201010116325 CN201010116325 CN 201010116325 CN 201010116325 A CN201010116325 A CN 201010116325A CN 102148993 B CN102148993 B CN 102148993B
Authority
CN
Grant status
Grant
Patent type
Prior art keywords
sub
band
module
wavelet
list
Prior art date
Application number
CN 201010116325
Other languages
Chinese (zh)
Other versions
CN102148993A (en )
Inventor
左雯
王宁
于培松
郭秀江
卓力
田卫
李晓光
周真理
张菁
Original Assignee
中兴通讯股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Grant date

Links

Abstract

本发明公开了一种小波图像的编码方法,包括:利用金字塔结构对经小波分解所得的各子带对应的小波系数进行分类,再对已分类的小波系数执行对应的编码操作,得到各子带的压缩码流;根据各子带压缩码流的率失真特性,将目标码率在各子带的压缩码流间进行最优分配,依据各子带被分配的码率对各子带的码流进行截断,得到对小波图像进行编码的码流。 The present invention discloses a method for encoding wavelet image, comprising: using a pyramid of wavelet coefficients of each sub-band obtained by the wavelet decomposition to the corresponding classification, and then performs an encoding operation corresponding to the wavelet coefficients classified, to give each subband the compressed stream; according to the distortion characteristic of each sub-band compressed code stream, compressed in each sub-optimal allocation of inter-stream target code rate band, each subband based on the code rate is allocated for each sub-band flow cut off, to obtain a code stream encoded image wavelet. 本发明还同时公开了一种小波图像的编码装置,运用该方法和装置可提高编码所得码流的抗误码能力,且码流支持多种空间分辨率的解码。 The present invention also discloses a wavelet image coding apparatus, using the method and apparatus can improve the error resilience of the encoded bit stream obtained, and support a variety of spatial resolution stream decoding.

Description

一种小波图像的编码方法和装置技术领域[0001] 本发明涉及数字图像编码和视频编码技术领域,尤其涉及一种小波图像的编码方法和装置。 Encoding method and apparatus in wavelet image Technical Field [0001] The encoding method and apparatus of the present invention relates to digital video coding and image coding technology, and in particular relates to an image wavelet. 背景技术[0002]自20世纪80年代以来,小波变换因其特有的与人眼视觉特性相符的多分辨率分析能力及方向选择能力,而被广泛应用于图像压缩领域,并取得了很好的效果。 [0002] Since the 1980s, the wavelet transform its analytical capabilities and multi-resolution direction-specific characteristics consistent with the human visual ability to choose, and is widely used in image compression, and achieved good effect. [0003] 图像经小波分解后可得到一系列不同分辨率的子带,不同分辨率的子带对应的频率也不同,之后再对得到的子带进行编码,也就是对小波图像进行编码,从而达到压缩图像的目的,以便执行后续图像传输的操作。 Decomposition [0003] The image obtained by the wavelet resolution of a number of different sub-bands, a frequency corresponding to the resolution of the different sub-bands are different, then after the obtained sub band coding, wavelet image is encoded so reaches the compressed object image so as to perform subsequent image transfer operation. 小波图像编码的关键就是如何更好地组织和表示小波系数,从而更好地利用小波系数所具有的统计特性,实现最大程度的数据压缩。 Wavelet image coding key is how to better organize and represent wavelet coefficients, wavelet coefficients in order to make better use of the statistical properties have achieved the greatest degree of data compression. 其中, 所述小波系数与图像经小波分解后所得的各子带对应,每个子带均有与之对应的多个小波系数。 Wherein each sub obtained after the wavelet coefficients by the wavelet transform with the image corresponding to each sub-band has a plurality of wavelet coefficients corresponding thereto. [0004]目前,小波图像编码的方法有很多,包括:矢量量化、标量量化、零树编码以及零块编码等,这些编码方法中,编码效率比较高的主要有两种:基于零树(Zero-tree)的小波编码方法和基于零块(Zero-block)的小波编码方法。 [0004] Currently, there wavelet image coding many ways, comprising: vector quantization, scalar quantization, zerotree coding and zero block coding, these coding methods, the encoding efficiency is high are mainly two: Zerotree (Zero -tree) wavelet coding and wavelet zero block coding based (zero-block) is. [0005] 其中,所述基于零树的小波编码方法中,比较有代表性的有嵌入式零树小波(EZW, Embedded Zero-tree Wavelet)编码方法和分层小波树集合分裂(SPIHT, Set Partitioning In Hierarchical Trees)编码方法。 [0005] wherein the wavelet coding method based on a zero tree, the more representative are EZW (EZW, Embedded Zero-tree Wavelet) coding method and a hierarchical tree wavelet set partitioning (SPIHT, Set Partitioning In Hierarchical Trees) coding method. 这里,所述EZW编码方法充分利用了不同尺度间小波系数的相似特性,有效地剔除了对高频小波系数的编码,极大地提高了小波系数的编码效率; 所述SPIHT编码方法,是在EZW编码方法基础上提出的一种更高效的小波图像编码方法,通过设计空间方向树来更有效地组织小波系数,所述空间方向树的数据结构不仅充分利用了不同尺度间小波系数的相关性,也充分考虑了同一尺度下小波系数的相关性,因此可以更有效地组织小波系数,提高编码效率。 Here, the full use of the EZW coding method similar characteristics between different scales of the wavelet coefficients, effectively eliminate the high frequency wavelet coefficients coding, coding efficiency is improved greatly wavelet coefficients; the coding method SPIHT, in EZW a more efficient wavelet image coding method based on the proposed encoding method, spatial orientation by designing more effectively organized tree wavelet coefficients, the spatial orientation data structure tree only make full use of the correlation between different scales of the wavelet coefficients, and taking into account the correlation of the wavelet coefficients at the same scale, it can be more effectively organized wavelet coefficients to improve coding efficiency. [0006] 所述基于零块的小波编码方法,采用四叉树分解的小波系数组织方式,即将图像分成四个区域,各区域对应一个集,如果这个集中的所有像素值都小于当前预设的阈值,则认为这个集是不重要,是“零块”,用比特O就可以表示该集中所有小波系数的状态;如果这个集中的所有像素值都大于当前预设的阈值,则认为这个集是重要的,将这个区域再分裂成四个子区域得到四个子集,然后对这四个子集重复进行重要性检测,对重要的子集仍然采用所述四叉树分解方法继续进行分裂,直到确定当前阈值下所有重要的小波系数。 [0006] wavelet coding method based on the zero block, wavelet coefficients organized quadtree decomposition, i.e. the image is divided into four regions, each region corresponding to a set, if the set of all the pixel values ​​are smaller than the current preset threshold, it is considered that set is not important, is the "zero blocks", with bits O can be represented in the set of all wavelet coefficients state; if the all pixel values ​​of set are greater than the current preset threshold value, it is considered that set is importantly, this area will again split into four sub-regions get four subsets, then these four subsets repeated the importance of testing for important subset still use the quad-tree decomposition method continues to divide until it is determined the current all important wavelet coefficients threshold. 基于零块的小波编码主要有集合分裂嵌入块编码方法(SPECK, Set Partitioned Embedded Block Coder)和优化截断点的嵌入块编码方法(EBC0T, Embedded Block Codingwith Optimized Truncation)等。 Set Partitioning Embedded Block Coding with a main wavelet coding method based on zero block (SPECK, Set Partitioned Embedded Block Coder) and Embedded Block Coding method optimized truncation point (EBC0T, Embedded Block Codingwith Optimized Truncation) and the like. [0007] 虽然现有的EZW、SPIHT和SPECK等小波编码方法具有嵌入的特性,但这些编码方法得到的嵌入式码流对误码非常敏感,由于所述已有的编码方法利用不同子带小波系数间的相关性,那么,如果码流的传输过程中一旦出现误码,小波系数间的相关性遭到破坏,则此误码之后的码流将失去同步,导致后续将码流解码出的图像“面目全非”,可见,现有小波编码方法的抗误码能力较差。 [0007] While the existing EZW, SPIHT, and the like SPECK embedded wavelet coding method having the characteristics, but the embedded coding code stream obtained by the method is very sensitive to error, because the conventional coding method using different wavelet subbands the correlation coefficient between, then the transmission process of destruction of the correlation between the error, the wavelet coefficients if the event stream, this stream will then lose synchronization error, leading to the subsequent code stream is decoded image "unrecognizable" visible, error resilience conventional wavelet coding method is poor. 此外,同样由于现有编码过程中小波系数间的相关性,使得现有的小波编码方法不支持多种空间分辨率解码,也就是说,现有的小波编码方法得到的压缩码流经解码后只能得到固定分辨率的解码图像,应用范围受到限制。 Furthermore, also due to the correlation between the conventional process of encoding the wavelet coefficients, wavelet coding such that the conventional method does not support multiple spatial resolution of the decoded, that is, after passing through a compressed code decoding method for the conventional wavelet coding obtained to obtain a decoded image only fixed resolution, its application range is limited. 发明内容[0008] 有鉴于此,本发明的主要目的在于提供一种小波图像的编码方法和装置,可提高编码所得码流的抗误码能力,且码流支持多种空间分辨率的解码。 [0008] In view of this, the main object of the present invention to provide a method and apparatus for encoding wavelet image can be improved error resilience code stream obtained by encoding, and the code stream decoding supports multiple spatial resolutions. [0009] 为达到上述目的,本发明的技术方案是这样实现的:[0010] 本发明提供了一种小波图像的编码方法,该方法包括:[0011] 对各子带的小波系数进行L级金字塔结构构建,所述L为整数,L的最大值为max {Iog^, Iogf !·,其中,所述w和H分别为当前子带的宽度和高度;[0012] 为当前子带设置重要像素列表LSP和不重要节点列表LIN,并初始化设置重要像素列表LSP和不重要节点列表LIN中的内容及比特平面系数η的初始值大小,其中,所述η为整数,其值为卜所述c(i,j)为当前子带的小波系数值;^ 」5[0013] 对所构建金字塔结构中的所有小波系数进行分类;[0014] 对分类得到的重要像素列表LSP中的小波系数执行细化操作;[0015] 判断当前比特平面系数η与I的大小关系,如果η大于等于1,则令η=η_1,并从返回执行所述分类操作、细化操作,直至执行所述当前比特平面系数η与I的 [0009] To achieve the above object, the technical solution of the present invention is implemented as follows: [0010] The present invention provides a method for encoding wavelet image, the method comprising: [0011] the wavelet coefficients of each sub-band level L construction of the pyramid structure, said L being an integer, L is the maximum value max {Iog ^, Iogf ·, wherein w and H are the width and height of the current sub-band;! [0012] important with the current sub-set LSP list of insignificant pixel and LIN node list, and initialization of setting the pixel important and unimportant LSP list LIN node list and the initial value of the size coefficient η of bit-planes, wherein the η is an integer having a value of the BU wavelet coefficient values ​​c (i, j) for the current sub-band; ^ "5 [0013] for all wavelet coefficients pyramid structure constructed in classification; [0014] performed on the wavelet coefficients significant pixels listing LSP classification obtained in refinement operation; [0015] Analyzing the current bit-plane coefficient of magnitude [eta] with the relationship I, if [eta] is greater than or equal to 1, then let η = η_1, and returns from the classification operation performed, detailed operation, performed until the current bit coefficient η of the plane and I 大小关系的判断操作;如果η小于1,则结束当前子带的小波系数完全编码过程,得到各子带的压缩码流;·[0016] 根据各子带压缩码流的率失真特性,将目标码率在各子带的压缩码流间进行最优分配,依据各子带被分配的码率对各子带的码流进行截断,得到对小波图像进行编码的码流。 Analyzing the operation magnitude relationship; if η is less than 1, the end of the wavelet coefficients complete the encoding process of the current sub-band to obtain each subband compressed stream; - [0016] The distortion characteristic of each sub-band compression rate of the code stream according to the target compression rate in each sub-band allocated optimally among streams, truncating the code stream according to the respective sub-band is allocated for each sub-band rate to obtain the encoded bit stream wavelet image. [0017] 其中,所述为当前子带设置重要像素列表LSP和不重要节点列表LIN时,设置一个重要像素列表LSP和L个不重要节点列表LIN。 [0017] wherein said belt is provided important and unimportant pixel LSP list LIN node list, list of LSP provided a significant pixel and an insignificant L LIN node list for the current sub. [0018] 其中,所述在各子带的压缩码流间分配目标码率,依据被分配的码率截断码流的过程为:[0019] 根据各子带的率失真特性确定各子带对应的最优截断点,根据所述的最优截断点对各子带的压缩码流进行截断。 [0018] wherein, in the process of each sub-band allocation target compression rate stream between the truncated bitstream based on the allocated rate as: [0019] determining for each sub-band corresponding to each sub-band according to rate distortion characteristics the optimal cut-off point, according to the optimal cut-off point of the compression of each sub-band code stream truncation. [0020] 本发明还提供了一种小波图像的编码装置,该装置包括:编码模块和截断模块; 其中,[0021] 所述编码模块,用于对各子带的小波系数进行L级金字塔结构构建,所述L为整数,L的最大值为maxi log;;,其中,所述W和H分别为当前子带的宽度和高度;[0022] 为当前子带设置重要像素列表LSP和不重要节点列表LIN,初始化设置重要像素列表LSP和不重要节点列表LIN中的内容及比特平面系数η的初始值大小,其中,所述η为整数,其值为l_logrX<l#n|)j所述c(i,j)为当前子带的小波系数值;对所构建金字塔结构"- -1 5中的所有小波系数进行分类,并对分类得到的重要像素列表LSP中的小波系数执行细化操作;[0023] 还用于判断当前比特平面系数η与I的大小关系,确定η大于等于I时,令η=η_1, 并从返回执行所述分类操作、细化操作,直至执行所述当前比特平面系数η与I的大 [0020] The present invention also provides a wavelet image coding apparatus, the apparatus comprising: an encoding module and a truncation module; wherein [0021] the encoding module, for performing the L-level pyramid of wavelet coefficients for each sub-band Construction of the L is an integer, L is a maximum value maxi log ;;, wherein W and H are the width and height of the current sub-band; [0022] provided with a list of significant pixels and insignificant for the current sub-LSP LIN node list, the initial setup contents list LSP important and unimportant pixel LIN node list and the initial value of the size coefficient η of bit-planes, wherein the η is an integer having a value of l_logrX <l # n |) j the c (i, j) is the wavelet coefficient values ​​of the current sub band; pyramid structure constructed of "- all wavelet coefficients classified in -15, and performs pixel thinning operations significant wavelet coefficients obtained classification list of LSP ; [0023] further configured to determine the current bit-plane when the magnitude relation between the coefficient [eta] and I, greater than or equal I [eta] is determined, so that η = η_1, and returns from the classification operation performed, detailed operation, performed until the current bit plane and I is a large coefficient η 关系的判断操作;确定η小于I时,结束当前子带的小波系数完全编码过程,得到各子带的压缩码流,并将所得的各子带的压缩码流传输给截断模块;[0024] 所述截断模块,用于根据编码模块所发的各子带压缩码流的率失真特性,将目标码率在各子带的压缩码流间进行最优分配,并依据各子带被分配的码率对各子带的码流进行截断,得到对小波图像进行编码的码流。[0025] 其中,所述编码模块进一步包括:金字塔构建模块、初始化模块、分类模块、细化模块和量化模块;其中,[0026] 所述金字塔构建模块,用于为当前编码的子带构建对应的金字塔结构,并将构建结果发送到初始化模块;[0027] 所述初始化模块,用于为当前子带设置重要像素列表LSP和不重要节点列表LIN, 并初始化设置不重要节点列表LIN和重要像素列表LSP中的内容及比特平面系数η的初始值 Analyzing the relationship between the operation; η is determined smaller than I, the end of the current sub-band coding process is complete wavelet coefficients, to obtain each sub-band of the compressed bit stream, the sub-code and the resulting compressed stream to cut the belt module; [0024] the truncation module is configured with a rate-distortion characteristics of the compressed stream according to the respective sub-coding module issued by the compression in each sub-optimal allocation of inter-band code stream target code rate, and is allocated according to each sub-band the code stream rate for each sub-band truncation to obtain a code stream encoded images based on [0025] wherein said encoding module further comprising: a pyramid construction module, initialization module, a classification module, refinement module and a quantization module ; wherein [0026] the pyramid building blocks used to construct the current sub-band coding corresponding pyramid, and send the results to build the initialization module; [0027] the initialization module is configured to set the current sub-band LSP list of important and unimportant pixel node list LIN, and initialization of setting is not important LIN node list and the list of significant pixels in the LSP coefficients and the bit-planes of the initial value of η 小,并将设置的结果发送到分类模块;[0028] 所述分类模块,用于以初始化模块所发的初始值为依据对当前子带对应的金字塔结构中的所有小波系数进行分类,并将分类结果发送到细化模块;还用于收到量化模块所发的当前比特平面系数η后,以新的η为依据继续执行分类操作;[0029] 所述细化模块,用于对分类模块分类得到的重要像素列表LSP中的小波系数执行细化操作,并在执行完当前细化操作后触发量化模块;还用于收到量化模块的通知后将当前子带的所有小波系数的细化结果发送到截断模块;[0030] 所述量化模块,用于判断分类模块中当前比特平面系数η与I的大小关系,如果η 大于等于1,则η减1,将递减后的η发送到分类模块,并触发分类模块以新的η为依据继续执行分类操作; [0031] 所述量化模块,还用于确定分类模块中当前比特平面系数η小 Small, and transmits the result set to the classification module; [0028] The classification module is configured to initialize the initial value of the send module based on the current pyramid structures all wavelet coefficients in subband classification, and classification results to refinement module; after receiving the quantization module is further configured to send the current bit-plane coefficient [eta], [eta] is based on a new classification operation to continue; [0029] the refinement module, a classification module important wavelet coefficients obtained by classifying the pixels in a listing LSP refinement operation is performed, and quantization module triggered after performing this refining operation; further used for all wavelet coefficients quantization module will be notified of the current sub-band refinement truncated result to the module; [0030] the quantization module configured to determine the classification module the current bit-plane coefficient η I and the magnitude relation, if η is greater than or equal to 1, the minus η 1, η will be sent down to the classification module, and a classification module triggers a new η classification based continue operation; [0031] the quantization module is further configured to determine the current bit-plane classification module small coefficient η I时,通知细化模块将当前子带的所有小波系数的细化结果发送到截断模块。 I, the notification will be sent refinement module to refine the results of all the wavelet coefficients to cutoff the current sub-band module. [0032] 上述方案中,所述截断模块在各子带的压缩码流间分配目标码率,依据被分配的码率截断码流的操作为:根据各子带的率失真特性确定各子带对应的最优截断点,根据所述的最优截断点对各子带的压缩码流进行截断。 [0032] In the above embodiment, the truncation module allocation between target code rate for each sub-band compressed stream, the stream truncation according to bit rate assigned: determining a distortion characteristic according to each sub-band of each sub-band corresponding to the optimal cut-off point, according to the optimal cut-off point of the compressed code stream truncating each subband. [0033] 本发明提供的小波图像的编码方法和装置,利用金字塔结构对经小波分解所得的各子带对应的小波系数进行分类,再对已分类的小波系数执行对应的编码操作,得到各子带的压缩码流;根据各子带压缩码流的率失真特性,将目标码率在各子带的压缩码流间进行最优分配,依据各子带被分配的码率对各子带的码流进行截断,所得码流即为对小波图像进行编码的码流。 [0033] The encoding method and apparatus of the present invention provides a wavelet image, using the pyramid structure by the wavelet decomposition wavelet coefficients of each sub resulting band corresponding to classify, then execute the code corresponding to the operation of the wavelet coefficients classified, to give the sub- with the compressed stream; distortion of each sub-band code stream compression rate depending on the characteristics, will be compressed in each sub-optimal allocation of inter-band code stream target code rate, code rate based on each subband assigned to each sub-band stream truncated, the resulting code stream is the images based on the stream encoding. 本发明的编码过程是分别对各子带独立进行编码,没有涉及到不同子带小波系数间的相关性,编码过程相对现有简单,因此,如果传输中某个子带的码流发生误码不会影响到其它子带的码流,抗误码能力得到提高。 The encoding process of the present invention is independent for each sub-band are encoded, not related to the correlation, the conventional encoding process is relatively simple, and therefore, if a transmission code stream generating subband error not different between the subbands of wavelet coefficients codestream will affect other sub-bands, improved error resilience. [0034] 此外,由于编码最终得到的码流是不同级数的子带码流排列组成的,因此,在解码时可有选择地对部分级数的码流进行解码,从而得到不同分辨率的解码图像,用户体验得到提闻。 [0034] In addition, since the resulting stream encoding different series arrangement of sub-band stream composition, therefore, may be selectively decoding the code stream decoding part series, whereby different resolutions a decoded image, the user experience mention smell. 附图说明[0035] 图1为本发明小波图像的编码方法实现流程示意图;[0036] 图2为本发明小波编码后所得的输出码流的结构示意图;[0037] 图3为本发明所述对各子带的小波系数进行完全编码的方法实现流程示意图;[0038] 图4为本发明所述金字塔结构组织方法构建所得的结构示意图;[0039] 图5为本发明小波图像编码的装置结构示意图。 BRIEF DESCRIPTION OF wavelet image coding method to achieve a schematic flow [0035] of the present invention, FIG. 1; the resulting structural diagram of the wavelet encoded output symbol stream [0036] FIG. 2 of the present invention; [0037] FIG. 3 of the present invention wavelet image coding apparatus configuration [0039] FIG. 5 of the present invention; a method for encoding a full wavelet coefficients of each sub-band to achieve a schematic flow diagram; [0038] Fig 4 a schematic view of the pyramid structures organizational structure of the resulting construction method of the present invention schematic diagram. 具体实施方式[0040] 本发明的基本思想是:利用金字塔结构对经小波分解所得的各子带对应的小波系数进行分类,再对已分类的小波系数执行对应的编码操作,得到各子带的压缩码流;根据各子带压缩码流的率失真特性将目标码率在各子带的压缩码流间进行最优分配,依据各子带被分配的码率对各子带的码流进行截断,所得码流即为对小波图像进行编码的码流。 DETAILED DESCRIPTION [0040] The basic idea of ​​the invention is: each sub pair wavelet-resulting decomposition using a pyramid structure with the wavelet coefficients corresponding to classify, then execute the code corresponding to the operation of the wavelet coefficients classified, to give each subband compressed stream; distortion characteristics of each sub-band based on the compression rate of the bit stream compressed in each sub-optimal allocation of inter-band code stream target code rate, the code stream for each subband based on each sub-band allocated rate truncated, the resulting code stream is the stream for encoding the wavelet image. [0041] 下面结合附图及具体实施例对本发明作进一步详细说明。 [0041] The following specific embodiments and examples of the present invention is described in further detail in conjunction with the accompanying drawings. [0042] 图1为本发明小波图像的编码方法实现流程示意图,如图1所示,该流程实现步骤如下:[0043] 步骤101 :利用金字塔结构对经小波分解所得的各子带对应的小波系数进行分类,再对已分类的小波系数执行对应的编码操作,得到各子带的压缩码流; [0042] Figure 1 is a wavelet image coding method of the invention to achieve a schematic flow diagram, shown in Figure 1, the process steps are as follows: [0043] Step 101: using a pyramid structure each sub-band obtained by the wavelet decomposition corresponds to the wavelet coefficient classification, then performing the operation corresponding to the encoding of wavelet coefficients classified, to obtain a compressed code stream of each sub-band;

[0044] 具体为:图像经小波分解后得到对应的小波图像,可根据需要将原始图像进行不同级数的分解,如:进行三级小波分解,分解过程为现有技术,此处不再详述;之后分别对经小波分解所得的各子带对应的小波系数执行编码操作,分别得到各子带的压缩码流。 [0044] Specifically: to obtain wavelet image corresponding to the image by the wavelet decomposition, the original image may be different stages of decomposition required, such as: for three decomposition, decomposition process of the prior art, not described in detail here described later; after an encoding operation are performed for each sub-band wavelet decomposition obtained by the wavelet coefficients corresponding to each sub-band respectively compressed stream. 其中,所述完全编码是指对各子带独立进行编码,对各子带的编码过程不涉及各子带小波系数间的相关性。 Wherein said means for fully coded independently for each sub-band, encoding process for each sub-band does not relate to the correlation between the wavelet coefficients of each sub-band. [0045] 这里,所述对各子带的小波系数进行完全编码的过程中,首先对各子带的小波系数进行I级金字塔结构构建,所述I为整数,I的最大值为maxMog1: JogD,其中,所述W 和H分别为当前子带的宽度和高度;并为每个子带设置一个重要像素列表(LSP,List of Significant Pixels)和I 个不重要节点列表(LIN, List ofInsignificant Node),每一级金字塔对应一个LIN,本发明中,将LSP描述为第一像素列表;将LIN描述为第二节点列表。 [0045] Here, the complete process of encoding the wavelet coefficients of each sub-band, first wavelet coefficients of each sub-band level pyramid structure constructed I, I is the integer, the maximum value of I is maxMog1: JogD , wherein W and H are the current sub-band width and height; set for each sub-band and a list of significant pixels (LSP, list of significant pixels) and the I unimportant node list (LIN, list ofInsignificant node) each level of the pyramid corresponds to a LIN, the present invention will be described as a first pixel LSP list; LIN will be described as a list of the second node. 对每个子带的小波系数进行编码时,首先要初始化当前子带的第一像素列表和所有第二节点列表,同时要初始化比特平面系数n,所述η为整数,其值为j logf—j,所述c(i,j)为当前子带的小波系数值;然后根据η对每一级金字塔结构中的系数进行分类,分类过程从金字塔最高级开始一直到最低级结束,得到第一像素列表,再对第一像素列表中的系数进行细化,得到每个系数的第η个最重要比特(MSB,Most SignificantBit);将n_l,重复上述的分类和细化过程,直到n=0为止,得到n+1个比特平面;最后,利用二进制算术编码来进一步提高压缩效率。 When coding the wavelet coefficients of each sub-band, a first pixel to be initialized and the list of the current sub-band a second list of all the points, and to initialize the bit-plane coefficient n, the η is an integer having a value of j logf-j the c (i, j) of wavelet coefficient values ​​of the current sub-band; and then classifying each coefficient in a pyramid structure according to [eta], the classification process is started from the highest level to the lowest level of the pyramid has been completed, to obtain a first pixel list, then the coefficient of the first pixel in the list of refinement, to obtain each coefficient η of most significant bits (MSB, most SignificantBit); the N_L, repeat the classification and the thinning process, up until n = 0 give the n + 1 bit-plane; finally, the use of binary arithmetic coding to further improve compression efficiency. [0046] 步骤102 :根据各子带压缩码流的率失真特性,将目标码率在各子带的压缩码流间进行最优分配,依据各子带被分配的码率对各子带的码流进行截断;[0047] 具体为:从各子带经编码所得的压缩码流中,利用现有的计算方法得到各子带的率失真特性;根据各子带对应的率失真特性,将目标码率在各子带的压缩码流间进行分配。 [0046] Step 102: the compression rate of the distortion in each sub-band according to the characteristics of the stream, the compression in each sub-optimal allocation of inter-band code stream target code rate, code rate based on each subband assigned to each sub-band truncated codestream; [0047] specifically: encoded with compressed stream obtained from each of the child, using conventional calculation method of each sub-band obtained distortion characteristic; the distortion characteristics of each sub-band corresponding to the distributed between the target code rate compressed stream for each sub-band. 这里,所述目标码率如何分配可以为:根据各子带的率失真特性确定各子带对应的最优截断点,使得各子带的截断码率之和小于目标码率时,整幅图像的编码失真率达到最小,也就达到了所述的最优分配。 How Here, the target bit allocation may be as follows: When determining the optimal truncation point for each sub-band corresponding to that of the cut rate for each sub-band is smaller than the target code rate and the each sub-band according to the rate-distortion characteristics of the entire image the minimum coding distortion rate, it reaches the optimal distribution. 这样,所述将目标码率在各子带的压缩码流间进行分配的过程,也就是在各子带的最优截断点处将各子带截断,截断后所得的截断码率即为各子带被分配的码率。 Thus, the target compression rate in each sub-band assignment between the process stream, i.e. at the optimal truncation point for each sub-band of each sub-band truncated, truncated obtained after each cut rate is the rate allocated subbands. [0048] 其中,由于各子带有多个截断点,因此需找到各子带的最优截断点;所述率失真特性是指子带被某一截断点截断后所对应的失真率,子带被任一截断点分割后所对应的失真率均可用现有的计算方法得到;所述截断码率是指各子带被截断点截断后的码率。 [0048] However, since each of the sub with a plurality of cut-off points, and therefore we need to find the optimal truncation point for each sub-band; refers to the rate-distortion characteristics of the sub-band distortion ratio is truncated after a certain cut-off point corresponding to the sub after any of the breakpoint with a length corresponding to the divided rate distortion calculation method obtained with available; refers to the runoff rate for each subband rate truncation point after being cut off. 上述目标码率分配的过程可用公式(I)表示如下:[0049] min D = D尸)s.七L =文if) ^Lmax (I)/-1 i=i[0050] 其中,所述D表示整幅图像的编码失真率,所述L表示各子带的截断码率之和, 所述N表示图像经小波分解后所得子带的数目,N与小波分解的级数ρ间的关系为:N = 3*p+l,例如:图像经三级小波分解后得到10个子带,Lmax为目标码车Lu ]mu 1分别表示第i个子带的截断点为Zi,i = 1,2,…,N时,该子带所对应的码率,即截断码率和失真率。 A target bit allocation procedure as described above using equation (I) is as follows:. [0049] min D = D corpse) L = S seven packets if) ^ Lmax (I) / - 1 i = i [0050] wherein said D represents the coding distortion of the entire image, and L represents the cut rate and the each sub-band, N represents the relationship between the number of stages resulting ρ subbands by wavelet decomposition of the image, the wavelet decomposition of N is: N = 3 * p + l, for example: an image 10 obtained by the three subbands of wavelet decomposition, Lmax target code cart Lu] mu 1 represent i th sub-band cut-off point for Zi, i = 1,2 , ..., N, the sub-band corresponding to the code rate, i.e., cut rate and distortion rate. [0051] 可以发现公式(I)是一个约束条件下的最优化问题,利用拉格朗日乘子法则可以将约束条件下的最优化问题转化为无约束的最优化问题,那么,引入拉格朗日乘子λ得到:N[0052] Ό + λ L =Yj (D;z·、+ /I · 4Z/}) ( 2 )f=][0053]设:J = Dfi] + λ . Lfi} (3 )[0054] V(X) = ^Lf0 ( 4)i=l[0055] 根据拉格朗日乘子法,在给定λ的情况下,使公式(3)达到最小,同时也满足公式(4)的截断点<,i = 1,2,···,Ν—定也是公式(I)的最优解,这样,搜索各子带的最优截断点ζ*, i = 1,2,…,N的过程,就简化为求解L' (λ) = Lmax的λ X值。 [0051] The equation can be found in (I) is an optimization problem under a constraint condition using the Lagrange multiplier rule can be optimized under constraints problem into an unconstrained optimization problem, then, lead Ru Lage Lagrangian multiplier [lambda] obtained: N [0052] Ό + λ L = Yj (D; z ·, + / I · 4Z /}) (2) f =] [0053] provided: J = Dfi] + λ Lfi. } (3) [0054] V (X) = ^ Lf0 (4) i = l [0055] the method of Lagrange multipliers, in the case where λ given in the equation (3) to a minimum, but also satisfies the equation (4) of the cut-off points <, i = 1,2, ···, Ν- is given formula (I) is the optimal solution, so that the optimal cutoff point search of each subband ζ *, i = 1 , 2, ..., N the process, reduces to solving L '(λ) = λ X value of Lmax. 因为Zi对应的是一些离散的采样点,因此本发明中采用二分法来搜索λχ值。 Because Zi corresponding to some discrete sampling points, thus the present invention is employed to search for λχ dichotomy value. 在给定λ的情况下,求解使公式(4)最小的<,i = 1,2,···,N的过程可以在每个子带内独立进行。 In the case where λ is given, so that to solve the formula (4) minimum <, i = 1,2, ···, N process can be carried out independently in each subband. 因此,对每个小波子带, 本发明提出了如下的快速搜索算法:[0056]初始化:设 Zopt=O ;[0057] for Zi=Ij 2, ···, Zi[0058]设M = Lfi) - Lfopt! AD = Dfi) - D]7、')[0059]如果 Λ D/ Λ L> λ,Zopt=Zi ;[0060]结束:[0061] 上述算法即为:设截断点的最优取值为z_,并令的初始值为0,之后对第i个子带的所有截断点依次执行AD/AL与λ的比较运算,确定AD/AL>X时,所得到的z_值则为截断点的最优取值,即得到所求的最优截断点Z:,之后在最优截断点Z; 位置处将各子带的码流截断,各子带经截断的码流形成最后的输出码流。 Thus, for each of the wavelet sub-band, the present invention proposes a fast search algorithm: [0056] Initialization: Let Zopt = O; [0057] for Zi = Ij 2, ···, Zi [0058] M = Lfi provided !) - Lfopt AD = Dfi) - D] 7, ') [0059] If Λ D / Λ L> λ, Zopt = Zi; [0060] end: [0061] the algorithm that is: set the optimal truncation point value is z_, and so when the initial value is 0, then successively performing AD / AL and comparison operations for all λ cut-off point of the i th sub-band, determining AD / AL> X, compared with a value obtained z_ the optimal cut point values, i.e., the optimal cut-off point required after :, Z-Z at the optimum cut-off points; flow formed last symbol stream at a cut position of each subband, each subband is truncated output stream. [0062] 图2为本发明小波编码后所得的输出码流的结构示意图,如图2所示,不同级数的子带依次排列输出,其中,所述LHUHLl和HHl用于标识一级小波分解后对应的子带,LH3、 HL3和HH3用于标识三级小波分解后对应的子带等等,图2中Ba等表示各子带被截断后对应的压缩码流,那么,如要对图2中的压缩码流进行解码,则可有选择地对压缩码流进行解码。 [0062] Fig 2 a schematic view of the structure of the present invention, the resulting wavelet-encoded output stream, as shown, a different number of stages of successively arranged sub-band output 2, wherein said LHUHLl HHl, and for identifying a wavelet decomposition the corresponding sub-band, LH3, HL3, and HH3 sub-band for the identity corresponding to three wavelet decomposition, etc., FIG. 2 and Ba represents each sub-band corresponding to the compressed stream is truncated so as to FIG. 2 compressed stream decoding can be selectively performed for decoding the compressed code stream. 假设原始图像的分辨率为512 X 512,如果选择LH3、HL3和HH3三个子带的压缩码流进行解码,那么所得图像的分辨率则为128 X 128 ;如果LH3、HL3、HH3、LH2、HL2和HH2六个子带的压缩码流进行解码,那么所得图像的分辨率则为256X256。 Assuming that the original resolution of the image is 512 X 512, if the selected compressed stream LH3, HL3 and HH3 three subband decoding, the resolution of the resulting image, compared with 128 X 128; if LH3, HL3, HH3, LH2, HL2 six and HH2 sub band compressed code stream is decoded, the resolution of the resulting image was 256X256. 当然,还可得到分辨率为512X512的图像,而现有的小波图像编码过程中涉及到各小波系数间的相关性,因此,现有技术对压缩码流进行解码后,只能得到分辨率与原始图像分辨率相同的图像,而不能得到其它多种与原始图像分辨率不同的图像。 Of course, also obtain an image resolution of 512X512, the conventional wavelet image coding process involves the correlation among the wavelet coefficients, and therefore, the prior art decoding the compressed code stream, and the only resolution is obtained the same image of the original image resolution, can not be obtained with a variety of other original images of different image resolution. [0063] 上述过程中,步骤101所述对各子带的小波系数进行完全编码的过程,如图3所示,每个子带的编码过程均包括金字塔结构的构建、初始化、分类、细化以及量化四个部分, 具体实现流程如下:[0064] 步骤301 :为当前子带构建对应的金字塔结构;[0065] 具体为:采用金字塔结构对当前子带对应的所有小波系数进行分类,图4所示为一个两级金字塔结构的构建方法,金字塔的所有第O级系数对应当前子带的所有小波系数;金字塔的第I级系数是通过对金字塔相应位置的4个第O级系数求最大值得到,其方法如公式(5)所示,以此类推,得到图4所示的金字塔结构。 [0063] The above-described process, the process of step 101 complete coding the wavelet coefficients of each sub-band, the encoding process shown in Figure 3 for each sub-band comprises a pyramid structure constructed, initialized, classification, refinement, and four quantization portion, the specific implementation process is as follows: [0064] step 301: Building a pyramid structure corresponding to the current sub-band; [0065] specifically: pyramid structure for all wavelet coefficients of the current sub-band corresponding to the classification, FIG. 4 illustrates a method for constructing two pyramid structure, all the coefficients of the first O-level pyramid wavelet coefficients corresponding to all the current sub band; class I pyramid four first coefficients by corresponding coefficient O stage of selecting the maximum value to obtain the position of the pyramids by a method as shown in formula (5), and so on, to give the pyramid structure shown in FIG 4. [0066] Cl(i, j)=max{c1_1(2i,2j), C1^1 (2i, 2 j + 1) , C1^1 (2i + l, 2 j) , 0^(21 + 1, 2j + l)}(5)[0067] 其中,I为整数,最大值为max{log【,logf},W和H分别为当前子带的宽度和高度; 所述C1Q, j)表示金字塔第I级(i, j)处的系数值,当I = O时,C1Q, j)对应的是当前子带的小波系数。 [0066] Cl (i, j) = max {c1_1 (2i, 2j), C1 ^ 1 (2i, 2 j + 1), C1 ^ 1 (2i + l, 2 j), 0 ^ (21 + 1, 2j + l)} (5) [0067] wherein, I is an integer and the maximum value {max {log, logf}, W and H are the current sub-band width and height; represents the pyramid of C1Q, j) class coefficient values ​​I (i, j) at when I = O, C1Q, j) is the wavelet coefficients corresponding to the current sub-band. [0068] 步骤302:为当前子带设置第一像素列表和第二节点列表,并初始化设置第二节点列表和第一像素列表中的内容及η的·初始值大小;[0069] 具体为:为当前子带设置一个第一像素列表和I个第二节点列表,LIN[1]表示当前子带的第I级金字塔系数对应的第二节点列表。 [0068] Step 302: a first sub-band set for the current node list and pixel list, and initializes the second set point list and the table of contents of the first pixel and the η-initial value of the size; [0069] Specifically: a sub-band set for the current pixel of the first list and second nodes list I, LIN [1] represents a second point of class I list pyramid coefficients of the current sub-band corresponding. 令金字塔结构中所有系数均为不重要系数,每一级金字塔结构中的系数归属于对应的LINk[l],此时第一像素列表则为空集,没有系数。 So that the pyramid structure of all coefficients are insignificant coefficients, each coefficient in a pyramid structure attributable to the corresponding LINK [L], while the first pixel listing null set, no coefficient. 本步骤所述初始化设置可表示为公式(6)、(7)和(8):[0070] LIN[I] = {第I级金子塔对应的所有系数} ; (6)[0071] LSP= Φ ; (7)[0072]»= Llog;-)]; (8)[0073] 步骤303 :对所构建金字塔结构中的所有小波系数进行分类;[0074] 如图4中所示黑色圆点为重要节点,也就是本发明中所述重要系数。 In this step, the initial settings may be expressed as equation (6), (7) and (8): [0070] LIN [I] = {Class I pyramid coefficients corresponding to all}; (6) [0071] LSP = Φ; (7) [0072] »= Llog ;-)]; (8) [0073] step 303: for all wavelet coefficients in pyramid structure constructed classifying; [0074] as shown by black circles in FIG. 4 important node, which is an important factor in the present invention. 这里,系数的重要性定义为公式(9):[0075] Here, the importance of the coefficient is defined as Equation (9): [0075]

Figure CN102148993BD00101

[0076] Sn(i, j)是n的函数,Sn(i,j)=l时,表明此时Sn(i,j)对应的小波系数为重要系数; sn(i, j)=0时,表明此时Sn(i,j)对应的小波系数为不重要系数。 [0076] Sn (i, j) is a function of n when Sn (i, j) = l, indicating that it is Sn (i, j) corresponding to the significant coefficients of wavelet coefficients; sn (i, j) = 0 when , indicating that Sn (i, j) corresponding insignificant coefficient wavelet coefficients. [0077] 具体分类过程为:[0078] I)从金字塔顶端开始,即第I级金字塔开始,对其小波系数进行重要性判断;[0079] 2)如果Sn(i, j)的值为0,则将对应的C1 (i, j)留在LIN[1]中;[0080] 3)如果Sn(i,j)的值为1,[0081] 如果1>0时,LIN[1]中对应的小波系数为C1 (i, j)-2n ;[0082] 如果1=0时,将对应的小波系数C0(i,j)加入到第一像素列表中,且LIN[1]中对应的小波系数为C1Q, j)-2n ;[0083] 4)如果当前级的金字塔中的小波系数全部分类结束,则令I = 1-1,即进入第1-1 级金字塔,重复上述2)步,直至1=0。 [0077] The classification process is specifically: [0078] I) starting from the top of the pyramid, the pyramid begins i.e. Class I, determines its importance wavelet coefficients; [0079] 2) If Sn (i, j) is 0 , then the corresponding C1 (i, j) remain in the LIN [1]; and [0080] 3) If Sn (i, j) is 1, [0081] If 1> 0, LIN [1] in wavelet coefficients corresponding to C1 (i, j) -2n; [0082] If 1 = 0, the corresponding wavelet coefficients C0 (i, j) is added to the list of the first pixel, the LIN and [1] corresponding the wavelet C1Q, j) -2n; [0083] 4) If the current level of the pyramid of wavelet coefficients in the classification of all the end, then let I = 1-1, i.e. into the first pyramid level 1-1, repeating the above step 2) until 1 = 0. [0084] 5)最终得到当前子带在比特平面系数为η时的第一像素列表。 [0084] 5) finally obtained a first list when the current pixel in the sub-band bit-plane coefficient η. [0085] 步骤304 :对分类得到的第一像素列表中的小波系数执行细化操作;[0086] 具体为:输出当前比特平面系数为η时的第一像素列表中每个小波系数的第η个MSB,然后清空第一像素列表,使之为空集。 [0085] Step 304: performing a thinning operation on the first pixel wavelet coefficients obtained in the sorted list; [0086] specifically to: output a first coefficient of the current bit-plane pixel η list when the first wavelet coefficients of each η a the MSB, then empty the list of the first pixel, so that the empty set. 所述各系数的第η个MSB的输出过程为现有技术,此处不再详述。 The output of each coefficient η MSB of the process of the prior art, not described in detail here. [0087] 本发明中,从比特平面系数《=^logr^w)l)j开始,执行分类和细化过程,直到η=0 为止结束。 [0087] In the present invention, the bit-plane from the coefficient "= ^ logr ^ w) l) j starts, performing classification and refinement process until η = 0 until the end. [0088] 步骤305 :判断当前比特平面系数η与I的大小关系,如果η大于等于1,则令η=η-1,并返回步骤303 ;如果η小于I,则执行步骤306 ;[0089] 该步骤即为对比特平面系数η执行量化操作。 [0088] Step 305: judging whether the current bit-plane coefficient [eta] magnitude relationship I, if [eta] is greater than or equal to 1, then let η = η-1, and returns to step 303; if [eta] is less than I, execute step 306; [0089] this step is the performed on the bit-plane quantization coefficients η operation. [0090] 其中,所述η大于等于I时,表明对当前子带的编码过程还未结束;所述η小于I 时,表明对当前子带的编码过程已结束,因为对各子带的编码过程是从η取最大值开始到η 为O结束的。 When [0090] wherein greater than or equal η I, indicate that the current sub-band coding process has not been completed; η is less than said I, to indicate the current sub-band coding process is completed, because each sub-band coding process from the start to the maximum value η η is O end. [0091] 步骤306 :当前子带的小波系数完全编码过程结束。 [0091] Step 306: the complete coding wavelet coefficients during the current sub-band end. [0092] 这里,当前子带完全编码过程结束后,还需对最终输出的码流执行常规的二进制算术熵编码,以便进一步提高压缩效率。 [0092] Here, the current sub-band coding process ends completely, the stream needs to perform a conventional binary arithmetic entropy encoding the output of the final code, in order to further improve the compression efficiency. [0093] 为实现上述方法,本发明还提供了一种小波图像编码的装置,如图5所示,该装置包括:编码模块和截断模块;其中,[0094] 所述编码模块,用于利用金字塔结构对经小波分解所得的各子带对应的小波系数进行分类,再对已分类的小波系数执行对应的编码操作,得到各子带的压缩码流,并将所得的各子带的压缩码流传输给截断模块;[0095] 所述截断模块,用于根据编码模块所发的各子带压缩码流的率失真特性,将目标码率在各子带的压缩码流间进行最优分配,并依据各子带被分配的码率对各子带的码流进行截断,得到对小波图像进行编码的码流。 [0093] To achieve the above-described method, the present invention also provides an apparatus for wavelet image coding, as shown in FIG. 5, the apparatus comprising: an encoding module and a truncation module; wherein [0094] the encoding module, for use pyramid wavelet coefficients for each sub-band wavelet decomposition obtained by the corresponding classification, and then performs an encoding operation corresponding to the wavelet coefficients classified, to give each subband compressed stream, each sub-band and the resulting compressed codes stream to truncation module; [0095] the truncation module is configured with a compression rate of each stream according to the encoding module issued by the sub-distortion characteristics, and in each sub-band of the compressed stream between the target code rate assignment optimally , and truncating the code stream according to the respective sub-band is allocated for each sub-band rate to obtain the encoded bit stream wavelet image. [0096] 所述编码模块利用金字塔结构对经小波分解所得的各子带对应的小波系数进行分类,再对已分类的小波系数执行对应的编码操作,具体为:[0097] 编码模块先为当前子带构建对应的金字塔结构,并为当前子带设置第一像素列表和第二节点列表,初始化设置第一像素列表和第二节点列表中的内容及比特平面系数η的初始值大小;对所构建金字塔结构中的所有小波系数进行分类,并对分类得到的第一像素列表中的小波系数执行细化操作;[0098] 还用于判断当前比特平面系数η与I的大小关系,确定η大于等于I时,令η=η_1, 并从所述执行分类操作步骤开始继续执行后续的编码过程;确定η小于I时,结束当前子带的小波系数完全编码过程。 Wavelet coefficient [0096] The encoding module by using a pyramid of wavelet decomposition of each sub-band obtained corresponding to the classification, then performing the operation corresponding to the encoding of wavelet coefficients classified, in particular: [0097] The first encoding module for the current the initial value of the size corresponding subband pyramid structure constructed and arranged for the current sub-band of the first list and the second pixel point list, a list of initial setting of the first pixel and the second node in the list and contents of the bit-plane coefficient η; of their construction of all the wavelet coefficients in pyramid classification, and performs pixel thinning operations first wavelet coefficients obtained in the sorted list; [0098] further configured to determine the current bit-plane coefficient η I and the magnitude relation, determined greater than η It is equal to I, so that η = η_1, and resumes execution from the subsequent encoding process step of performing classification operation; determining [eta] is less than I, completely ending the encoding process the wavelet coefficients of the current sub-band. [0099] 所述编码模块进一步包括:金字塔构建模块、初始化模块、分类模块、细化模块和量化模块;其中,[0100] 所述金字塔构建模块,用于为当前编码的子带构建对应的金字塔结构,并将构建结果发送到初始化模块;[0101] 所述初始化模块,用于为当前子带设置第一像素列表和第二节点列表,并初始化设置第二节点列表和第一像素列表中的内容及比特平面系数η的初始值大小,并将设置的结果发送到分类模块;[0102] 所述分类模块,用于以初始化模块所发的初始值为依据对当前子带对应的金字塔结构中的所有小波系数进行分类,并将分类结果发送到细化模块;还用于收到量化模块所发的当前比特平面系数η后,以新的η为依据继续执行分类操作;[0103] 所述细化模块,用于对分类模块分类得到的第一像素列表中的小波系数执行细化操作,并在执行完当前 [0099] The encoding module further comprising: pyramid construction module, initialization module, a classification module, refinement module and a quantization module; wherein [0100] the pyramid construction module, for encoding the pyramid for the current sub-band corresponding to construct structure, and the results are sent to build the initialization module; [0101] the initialization module is configured with a first pixel and the second node list for the current sub-list, and initializes the list of the second node and the first set of pixels in the list content bit-planes and the initial value of the size coefficient η, and transmits the set to the result of the classification module; [0102] the classification module is configured to initialize the module to send an initial value based on the current sub-band corresponding to the pyramid structure of classify all wavelet coefficients, and sends the result to the classification refinement module; after receiving the quantization module is further configured to send the current bit-plane coefficient [eta], [eta] is based on a new classification operation to continue; [0103] the refinement module, for performing a thinning operation of the first wavelet coefficients pixel classification module list obtained in the classification, and executing the current 化操作后触发量化模块;还用于收到量化模块的通知后将当前子带的所有小波系数的细化结果发送到截断模块;[0104] 所述量化模块,用于判断分类模块中当前比特平面系数η与I的大小关系,如果η 大于等于1,则η减1,将递减后的η发送到分类模块,并触发分类模块以新的η为依据继续执行分类操作; [0105] 所述量化模块,还用于确定分类模块中当前比特平面系数η小于I时,通知细化模块将当前子带的所有小波系数的细化结果发送到截断模块。 After the operation trigger quantization module; further refinement for the results of all the wavelet coefficients after quantization module notified of the current sub-band to cut off transmission module; [0104] The quantization module configured to determine the current bit classification module plane coefficient η I and the magnitude relation, if η is greater than or equal to 1, the minus η 1, η transmitting the decremented to the classification module, and a classification module triggers a new η classification based continue operation; [0105] the the quantization module is further configured to determine the classification module the current bit-plane coefficient η is smaller than I, the notification will be sent refinement module to refine the results of all the wavelet coefficients to cutoff the current sub-band module. [0106] 以上所述,仅为本发明的较佳实施例而已,并非用于限定本发明的保护范围,凡在本发明的精神和原则之内所作的任何修改、等同替换和改进等,均应包含在本发明的保护范围之内。 [0106] The above are only preferred embodiments of the present invention but are not intended to limit the scope of the present invention, any modifications within the spirit and principle of the present invention, equivalent replacements and improvements are It is included within the scope of the present invention.

Claims (6)

  1. 1. 一种小波图像的编码方法,其特性在于,该方法包括:对各子带的小波系数进行L级金字塔结构构建,所述L为整数,L的最大值为max {log:, log;7},其中,所述W和H分别为当前子带的宽度和高度;为当前子带设置重要像素列表LSP和不重要节点列表LIN,并初始化设置重要像素列表LSP和不重要节点列表LIN中的内容及比特平面系数η的初始值大小,其中,所述η为整数,其值为卜所述c(i,j)为当前子带的小波系数值;对所构建金字塔结构中的所有小波系数进行分类;对分类得到的重要像素列表LSP中的小波系数执行细化操作;判断当前比特平面系数η与I的大小关系,如果η大于等于1,则令η=η-1,并返回执行所述分类操作、细化操作,直至执行所述当前比特平面系数η与I的大小关系的判断操作; 如果η小于1,则结束当前子带的小波系数完全编码过程,得到各子 CLAIMS 1. A method for encoding wavelet image, wherein the characteristics thereof, the method comprising: the wavelet coefficients of each sub-band L-level pyramid structure constructed, the L being an integer, L is the maximum value max {log :, log; 7}, wherein W and H are the current sub-band width and height; sub-band set as the current list of significant pixels and insignificant LSP LIN node list, and initializes the list of LSP provided significant pixels and insignificant in the node list LIN the contents of bit-planes and the initial value of the size coefficient η, wherein η is an integer having a value Bu said c (i, j) is the wavelet coefficient values ​​of the current sub band; all wavelet pyramid structure constructed classifying coefficient; thinning operations performed on the list of wavelet coefficients significant pixels obtained by classifying the LSP; determining the current bit-plane coefficient [eta] and I is the magnitude relation, if [eta] is greater than or equal to 1, then let η = η-1, and return execution the classification operation, detailed operation, performed until the current bit-plane coefficient η I and the magnitude relationship determination operation; if η is less than 1, the encoding process ends completely wavelet coefficients of the current sub-band, to give the sub- 带的压缩码流;根据各子带压缩码流的率失真特性,将目标码率在各子带的压缩码流间进行最优分配,依据各子带被分配的码率对各子带的码流进行截断,得到对小波图像进行编码的码流。 With the compressed stream; distortion of each sub-band code stream compression rate depending on the characteristics, will be compressed in each sub-optimal allocation of inter-band code stream target code rate, code rate based on each subband assigned to each sub-band truncated code stream to obtain a stream encoded image wavelet.
  2. 2.根据权利要求1所述的小波图像的编码方法,其特性在于,所述为当前子带设置重要像素列表LSP和不重要节点列表LIN时,设置一个重要像素列表LSP和L个不重要节点列表LIN。 The encoding method according to claim 1 wavelet image, wherein the characteristics of the band setting important and unimportant pixel LSP list node list LIN, provided a significant pixel and the L unimportant LSP list for the current sub-nodes list LIN.
  3. 3.根据权利要求1或2所述的小波图像的编码方法,其特性在于,所述在各子带的压缩码流间分配目标码率,依据被分配的码率截断码流的过程为:根据各子带的率失真特性确定各子带对应的最优截断点,根据所述的最优截断点对各子带的压缩码流进行截断。 The encoding method of claim 12 or claim wavelet image, the characteristics that the compression allocation in each sub-band between the target code rate streams, cut rate are assigned according to a process stream is: determining characteristics of each sub-band corresponding to the optimal truncation point, truncate each sub-band bit stream compressed according to the optimal cut-off point of the respective sub-bands according to the rate-distortion.
  4. 4. 一种小波图像的编码装置,其特性在于,该装置包括:编码模块和截断模块;其中, 所述编码模块,用于对各子带的小波系数进行L级金字塔结构构建,所述L为整数,L的最大值,其中,所述W和H分别为当前子带的宽度和高度;为当前子带设置重要像素列表LSP和不重要节点列表LIN,初始化设置重要像素列表LSP和不重要节点列表LIN中的内容及比特平面系数η的初始值大小,其中,所述η为整数,其值为卜L 」5所述c(i,j)为当前子带的小波系数值;对所构建金字塔结构中的所有小波系数进行分类, 并对分类得到的重要像素列表LSP中的小波系数执行细化操作;还用于判断当前比特平面系数η与I的大小关系,确定η大于等于I时,令η=η-1,并返回执行所述分类操作、细化操作,直至执行所述当前比特平面系数η与I的大小关系的判断操作;确定η小于I时,结束当前 A wavelet image coding apparatus, wherein the characteristics thereof, the apparatus comprising: an encoding module and a truncation module; wherein the coding module, for performing the L-level pyramid of wavelet coefficients of each sub-band constructed, the L is an integer, L is a maximum value, wherein W and H are the current sub-band width and height; provided with a list of significant pixels and insignificant LSP sub-list for the current node LIN, initialization setting is not important and significant pixel LSP list LIN node list and the contents of bit-planes coefficient η size of the initial value, wherein the η is an integer having a value of L Bu "5 c (i, j) is the wavelet coefficient values ​​of the current sub band; of their when further configured to determine the current bit-plane coefficient η I with magnitude relation, greater than or equal to determine η I; all wavelet coefficients in pyramid structure constructed classification, and performs pixel thinning operations significant wavelet coefficients obtained classification list of LSP , so that η = η-1, and returns to perform the classification operation, detailed operation, performed until the current bit-plane coefficient of magnitude [eta] with the relationship I determination operation; determining [eta] is less than I, the end of the current 带的小波系数完全编码过程,得到各子带的压缩码流,并将所得的各子带的压缩码流传输给截断模块;所述截断模块,用于根据编码模块所发的各子带压缩码流的率失真特性,将目标码率在各子带的压缩码流间进行最优分配,并依据各子带被分配的码率对各子带的码流进行截断,得到对小波图像进行编码的码流。 Completely with the encoding process wavelet coefficients of each subband to obtain compressed stream, each sub-code and the resulting compressed stream to cut the belt module; the truncation module is configured to compress according to the respective sub-band coding module issued by stream rate distortion characteristics of the target code rate in each sub-band compression optimal distribution between streams, and truncating the code stream according to the respective sub-band is allocated for each sub-band rate to obtain images based on coding code stream.
  5. 5.根据权利要求4所述的小波图像的编码装置,其特性在于,所述编码模块进一步包括:金字塔构建模块、初始化模块、分类模块、细化模块和量化模块;其中,所述金字塔构建模块,用于为当前编码的子带构建对应的金字塔结构,并将构建结果发送到初始化模块;所述初始化模块,用于为当前子带设置重要像素列表LSP和不重要节点列表LIN,并初始化设置不重要节点列表LIN和重要像素列表LSP中的内容及比特平面系数η的初始值大小,并将设置的结果发送到分类模块;所述分类模块,用于以初始化模块所发的初始值为依据对当前子带对应的金字塔结构中的所有小波系数进行分类,并将分类结果发送到细化模块;还用于收到量化模块所发的当前比特平面系数η后,以新的η为依据继续执行分类操作;所述细化模块,用于对分类模块分类得到的重要像素列表L The encoding device according to claim wavelet image, which is characteristic in that, said encoding module further comprising: pyramid construction module, initialization module, a classification module, refinement module and a quantization module; wherein the pyramid construction module used for the current sub-band coding construct corresponding pyramid, and send the results to build the initialization module; the initialization module is provided with a list of significant pixels and insignificant LSP as the current sub-node list LIN, and initialization settings LIN node list is not important and significant pixel content in the LSP list and the initial value of the size coefficient η of bit-planes, and transmits the result set to the classification module; the classification module is configured to initialize the module made according to the initial value of All pyramid wavelet coefficients corresponding to the current sub-band are classified, and sends the result to the classification refinement module; after receiving the quantization module is further configured to send the current bit-plane coefficient [eta], [eta] is a new basis continued performs classification operation; the refinement module, a classification module to classify the pixel important to obtain a list L SP中的小波系数执行细化操作,并在执行完当前细化操作后触发量化模块;还用于收到量化模块的通知后将当前子带的所有小波系数的细化结果发送到截断模块;所述量化模块,用于判断分类模块中当前比特平面系数η与I的大小关系,如果η大于等于1,则η减1,将递减后的η发送到分类模块,并触发分类模块以新的η为依据继续执行分类操作;所述量化模块,还用于确定分类模块中当前比特平面系数η小于I时,通知细化模块将当前子带的所有小波系数的细化结果发送到截断模块。 Wavelet coefficients SP performs thinning operation, and trigger quantization module after performing this refining operation; refine the results of all the wavelet coefficients after the quantization module is further configured to receive notification of the current sub-band to cut off the transmission module; the quantization module configured to determine the classification module the current bit-plane coefficient η I magnitude relation, if [eta] is greater than or equal to 1, the minus 1 [eta], [eta] will be sent to the classification module decremented, and trigger a new classification module η continue operations classified based; the quantization module is further configured to determine the classification module when the current bit-plane coefficient η is less than I, the notification will be sent refinement module to refine the results of all the wavelet coefficients to cutoff the current sub-band module.
  6. 6.根据权利要求4或5所述的小波图像的编码装置,其特性在于,所述截断模块在各子带的压缩码流间分配目标码率,依据被分配的码率截断码流的操作为:根据各子带的率失真特性确定各子带对应的最优截断点,根据所述的最优截断点对各子带的压缩码流进行截断。 The encoding apparatus of claim 5 or claim wavelet image, the characteristics that the cutoff between the module allocation in the target code rate for each sub-band compressed stream, the stream truncation rate based on the allocated to: determine for each sub-band corresponding to the optimal truncation point, according to the optimal cut-off point of the compressed code stream for each sub-band truncation each sub-band according to the rate-distortion characteristics.
CN 201010116325 2010-02-10 2010-02-10 Method and device for encoding wavelet image CN102148993B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201010116325 CN102148993B (en) 2010-02-10 2010-02-10 Method and device for encoding wavelet image

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201010116325 CN102148993B (en) 2010-02-10 2010-02-10 Method and device for encoding wavelet image

Publications (2)

Publication Number Publication Date
CN102148993A true CN102148993A (en) 2011-08-10
CN102148993B true CN102148993B (en) 2013-03-27

Family

ID=44422953

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201010116325 CN102148993B (en) 2010-02-10 2010-02-10 Method and device for encoding wavelet image

Country Status (1)

Country Link
CN (1) CN102148993B (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6137915A (en) 1998-08-20 2000-10-24 Sarnoff Corporation Apparatus and method for error concealment for hierarchical subband coding and decoding
US20020080408A1 (en) 1999-12-17 2002-06-27 Budge Scott E. Method for image coding by rate-distortion adaptive zerotree-based residual vector quantization and system for effecting same

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6137915A (en) 1998-08-20 2000-10-24 Sarnoff Corporation Apparatus and method for error concealment for hierarchical subband coding and decoding
US20020080408A1 (en) 1999-12-17 2002-06-27 Budge Scott E. Method for image coding by rate-distortion adaptive zerotree-based residual vector quantization and system for effecting same

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
Ya-Qin Zhang,etl.Motion-Compensated Wavelet Transform Coding for Color Video Compression.《IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY》.1992,第2卷(第3期),
丁素英.基于小波变换的Laplacian金字塔图像数据压缩.《潍坊学院学报》.2009,第9卷(第4期),
党百振,等.数字图象的金字塔型预测编码.《桂林电子工业学院学报》.2000,第20卷(第1期),
吴宗泽,等.JPEG2000实时截断码率控制新算法及其VLSI结构设计.《电子学报》.2006,(第8期),

Also Published As

Publication number Publication date Type
CN102148993A (en) 2011-08-10 application

Similar Documents

Publication Publication Date Title
Tang et al. Three-dimensional wavelet-based compression of hyperspectral images
US6728316B2 (en) Video coding method
Da Silva et al. A successive approximation vector quantizer for wavelet transform image coding
US20050201624A1 (en) Method, program and apparatus for image processing capable of effectively performing image transmission, and a medium storing the program
US20020131497A1 (en) Apparatus and method for image coding using tree-structured quantization based on wavelet transform
US6519284B1 (en) Encoding method for the compression of a video sequence
US6049630A (en) Data compression using adaptive bit allocation and hybrid lossless entropy encoding
US20080037880A1 (en) Scalable, progressive image compression and archiving system over a low bit rate internet protocol network
Usevitch A tutorial on modern lossy wavelet image compression: foundations of JPEG 2000
Sudhakar et al. Image compression using coding of wavelet coefficients–a survey
US7356085B2 (en) Embedded multiple description scalar quantizers for progressive image transmission
Chew et al. Survey of image compression algorithms in wireless sensor networks
US6327392B1 (en) Method of visual progressive coding
US20080094259A1 (en) Context-Based Encoding and Decoding of Signals
EP0914004A1 (en) Coding system and method for lossless and lossy compression of still and motion images
US5923785A (en) System and method for compressing data
Nister et al. Lossless region of interest coding
Kharate et al. Color image compression based on wavelet packet best tree
US6549673B1 (en) Wavelet-based compression of images for storage, transmission and reconstruction using hierarchical subband decomposition
Algazi et al. Analysis-based coding of image transform and subband coefficients
JP2002034043A (en) Method and device for coding and recording medium
Akter et al. A modified-set partitioning in hierarchical trees algorithm for real-time image compression
US6625217B1 (en) Constrained wavelet packet for tree-structured video coders
US7068848B2 (en) Method of coding image data using region of interest
US6795505B2 (en) Encoding method for the compression of a video sequence

Legal Events

Date Code Title Description
C06 Publication
C10 Request of examination as to substance
C14 Granted