[go: up one dir, main page]

CN102314881B - MP3 (Moving Picture Experts Group Audio Layer 3) watermarking method for improving watermark-embedding capacity in MP3 file - Google Patents

MP3 (Moving Picture Experts Group Audio Layer 3) watermarking method for improving watermark-embedding capacity in MP3 file Download PDF

Info

Publication number
CN102314881B
CN102314881B CN2011102660642A CN201110266064A CN102314881B CN 102314881 B CN102314881 B CN 102314881B CN 2011102660642 A CN2011102660642 A CN 2011102660642A CN 201110266064 A CN201110266064 A CN 201110266064A CN 102314881 B CN102314881 B CN 102314881B
Authority
CN
China
Prior art keywords
watermark
codeword
file
value area
code word
Prior art date
Legal status (The legal status 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 status listed.)
Expired - Fee Related
Application number
CN2011102660642A
Other languages
Chinese (zh)
Other versions
CN102314881A (en
Inventor
杨小澎
陈真勇
毛佳俊
熊璋
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beihang University
Original Assignee
Beihang University
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
Application filed by Beihang University filed Critical Beihang University
Priority to CN2011102660642A priority Critical patent/CN102314881B/en
Publication of CN102314881A publication Critical patent/CN102314881A/en
Application granted granted Critical
Publication of CN102314881B publication Critical patent/CN102314881B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Editing Of Facsimile Originals (AREA)

Abstract

本发明提供一种用于提高MP3文件中水印嵌入容量的MP3水印方法，该方法基于小值区Huffman表码字与水印映射关系，包括码字分类、码字与水印映射、水印嵌入过程和水印提取过程；其中，水印嵌入之前，对MP3编码过程中使用的小值区Huffman表中的码字按照码字汉明重量不同进行分类，然后按照分类结果与汉明距离最小原则建立码字和水印间的映射关系；嵌入过程中，依次对MP3文件中的每帧数据进行解码，在解码到小值区时，每解码出一个小值区四元组值，就按照码字与水印间的映射关系嵌入水印，然后将修改后的码字写入MP3文件；最后，将所有处理完毕的MP3帧重构成含水印的MP3文件；本发明能在保证音频质量的同时，很大程度上提高MP3文件中的水印嵌入容量。

Figure 201110266064

The present invention provides an MP3 watermark method for increasing the capacity of watermark embedding in MP3 files. The method is based on the small value area Huffman table codeword and watermark mapping relationship, including codeword classification, codeword and watermark mapping, watermark embedding process and watermark Extraction process; wherein, before the watermark is embedded, the codewords in the small-value area Huffman table used in the MP3 encoding process are classified according to the codeword Hamming weight, and then the codewords and watermarks are established according to the principle of the minimum distance between the classification result and the Hamming distance During the embedding process, each frame of data in the MP3 file is decoded in turn. When decoding to the small-value area, each time a quadruple value in the small-value area is decoded, it will follow the mapping between the codeword and the watermark. Relationally embed watermark, then write the modified codeword into MP3 file; finally, reconstruct all processed MP3 frames into watermarked MP3 file; the present invention can greatly improve MP3 file while ensuring audio quality. The watermark embedding capacity in .

Figure 201110266064

Description

A kind of MP3 water mark method for improving mp3 file watermark embedding capacity
Technical field
The present invention relates to a kind of MP3 watermarking method, particularly a kind of MP3 water mark method for improving mp3 file watermark embedding capacity, the method is based on little value district Huffman table code word and watermark mapping relations.
Background technology
The numerical information revolution has brought deep variation for human society and life, has also brought new challenge and innovation opportunity simultaneously.The digital multimedia works can perfect copy with world wide in need hardly the distribution of cost, make the safety problems such as digital copyright protecting and content integrity checking become problem in the urgent need to address.The AIFC of moving at Internet and various machine at present has a lot, and the main audio file formats of two classes is wherein arranged: lossless format with diminish form.Lossless format comprises: WAV, PCM, TTA, FLAC, AU, APE, TAK, WavPack (WV) etc.; Diminishing form comprises: MP3, and Windows Media Audio (WMA), Ogg Vorbis (OGG), AAC etc., the most general with the audio file of MP3 compressed format in numerous audio file formats, spread scope also is the widest.The MP3 full name is MPEG-1Audio Layer III, it is the 3rd layer concrete application in the MPEG-1 audio standard, it can make ratio of compression reach 1: 10 or 1: 12 (L.G.Zhang under the condition that keeps CD Quality, R.D.Wang, and D.Q.Yan.Data hiding in mp3 audio by modifying qmdct coefficients[A] .In Computing, Communication, Control, and Management, 2009.CCCM2009.ISECS International Colloquium on; C.T.Wang, T.S.Chen, and W.H.Chao.A new audio watermarking based on modified discrete cosine transform of mpeg/audio layer iii[A] .In Networking, Sensing and Control, 2004 IEEE International Conference on), greatly save storage space, thereby also become more and more record companies and the music making people selects it as the first-selected audio format that is published on the internet.Although proposed again afterwards the audio formats such as AAC (audio scheme that proposes among the Advanced Audio Coding:MPEG-2), WMA, Ogg, VQF, they all increase on ratio of compression and audio quality, all fail to shake the dominant position of MP3.Under the restriction of consumer's factor and computer technology factor, MP3 will be audio format the most popular on the internet within the quite a long time.Yet; the just popularity of MP3 and convenience; so that the piracy of MP3 musical works and illegal download all become very frequent, this causes the legitimate rights and interests of creator and affiliated company thereof to be subject to serious infringement, and the MP3 Copyright Protection on the internet becomes the focus that receives much concern day by day.And digital watermarking with some Information hiding in digital picture, video or sound signal, become a kind of effective means of protecting digital media content safety.
Digital watermarking algorithm for MP3 is divided into following a few class according to the difference on opportunity that embeds at present: 1) the robust audio watermark of the compressed and decompressed attack of anti-MP3, such watermarking algorithm mainly is to carry out behind the time-frequency conversion embed watermark on frequency domain for the sound signal of original PCM, even its effect is original audio attacked by the MP3 compression and decompression after, still can extract watermark and carry out the functions such as copyright authentication, method (J.Wang such as people such as Wang, R.Healy, and J.Timoney.A robust audio watermarking scheme based on reduced singular value decomposition and distortion removal[J] Signal Processing, 2011.).2) in conjunction with the MP3 watermark of cataloged procedure, this type of watermarking algorithm mainly is the compression encoding process in conjunction with MP3, come embed watermark by the MDCT coefficient before the modification quantification or the MDCT coefficient after the quantification, or utilize other characteristics in the MP3 cataloged procedure to come embed watermark, its effect is that the MP3 watermarking algorithm has certain robustness, owing to self just combining the MP3 cataloged procedure, so compression and decompression have robustness to MP3, method (Akinori Ito etc. " Data Hiding is a Better Way for Transmitting Side Information for MP3 Bitstream " such as Akinori, 5th International Conference on Intelligent Information Hiding and Multimedia Signal Processing, 2009.).3) in conjunction with the MP3 watermark of decode procedure, this type of watermarking algorithm mainly is that mp3 file is carried out partial decoding of h, then come embed watermark by the partial data of revising in the MP3 code stream, the data of revising are generally large value district data, scale factor and little value district data etc., be characterized in that real-time is good, such as method (Rangding Wang, the etc. " A Novel data hiding algorithm for MP3 audio " of Wang, ICSP, 2008.)
Because MP3 mainly propagates at Internet, its popularity and velocity of propagation have determined that the MP3 watermark needs requirement of real time, therefore more can satisfy the requirement of real-time than the algorithm of other types in conjunction with the watermarking algorithm of MP3 decoding process.The people such as Yan have proposed a kind of improvement algorithm (Y.Diqun for the hidden writing tools of MP3Stego in 2009, W.Rangding, and Z.Liguang.Quantization step parity-based steganography for mp3 audio[J] .Fundamenta Informaticae, 2009,97 (1): 1-14), eliminated potential endless loop in the MP3Stego algorithm; Yan and Wang proposed in 2011 a kind of large value district Huffman table replace the MP3 watermark selected (D.Yan and R.Wang.Huffman table swapping-based steganograpy for mp3 audio[J] .Multimedia Tools and Applications, 2011, pages 1-15.), its algorithm on capacity more in the past the MP3 watermarking algorithm improve a lot, and guaranteed audio quality.The algorithm of Yan and Wang all is the watermarking algorithm in conjunction with the MP3 cataloged procedure, although its audio quality or capacity have good effect, real-time remains further to be improved, the space that also is improved on embedding capacity simultaneously.
Summary of the invention
The technical problem to be solved in the present invention is: overcome the deficiencies in the prior art, a kind of MP3 water mark method for improving mp3 file watermark embedding capacity is provided, the method is by the characteristic of code word in the little value of the research district Huffman table, use the method for code word classification to set up the mapping relations of code word and watermark, and utilize this pass to tie up to embed watermark in the mp3 file.This method can embed a large amount of watermarks in the situation that introduce less distortion, and its watermark embedding capacity method more in the past improves a lot.
The technical solution adopted for the present invention to solve the technical problems: a kind of MP3 water mark method for improving mp3 file watermark embedding capacity, the method comprises code word classification, code word and watermark mapping, watermark embed process and watermark extraction process; Described code word classification is that all code words in the little value district Huffman table are divided according to the Hamming weight difference; Described code word and watermark mapping is to set up mapping relations between code word and the watermark according to code word classification results and Hamming distance minimum principle; Described watermark embed process is: before watermark embeds, code word in the little value district Huffman that uses in the MP3 cataloged procedure table is classified according to code word Hamming weight difference, then set up mapping relations between code word and watermark according to classification results and Hamming distance minimum principle; In telescopiny, successively the every frame data in the mp3 file are decoded, when being decoded to little value district, whenever decode a little value district quaternary class value, just according to the mapping relations embed watermark between code word and watermark, then amended code word is write mp3 file; At last, the MP3 frame that all is disposed reconstitutes the mp3 file that contains watermark; Described watermark extraction process is: equally the every frame data in the mp3 file are decoded, when being decoded to little value district data, whenever decoding a little value district quaternary class value and just extract watermark according to the mapping relations between code word and watermark;
Described watermark embed process is specific as follows:
A. read in mp3 file, it is carried out the MP3 decoding operation.Decode procedure carries out according to the decoding process in the MP3 standard, and each frame data in the mp3 file after the large value district data of having decoded, are carried out the decode operation in little value district;
When b. being decoded to little value district data, whenever decoding a little value district four-tuple, find out target code word behind the embed watermark according to the little value district Huffman shown in the table 1 table code word and watermark mapping relations, this target code word will be replaced the corresponding original code word of four-tuple.If the corresponding code word of the four-tuple that decodes is the untapped code word in the table 1, then keep its code word constant;
C. after the little value district data in this frame being disposed according to step b, jointly reconstituting new frame data with other data of large value district's data and present frame, and jump to the processing that step b carries out next frame;
D. after all frames in the mp3 file all pass through the processing of watermark embedding operation, new MP3 frame is reconstituted the mp3 file that contains watermark.
Described watermark extraction process is specific as follows:
A1. read in mp3 file, it is carried out the MP3 decoding operation.Carry out according to the decoding process in the MP3 standard in the decode procedure, each frame data in the mp3 file after the large value district data of having decoded, are carried out the decode operation in little value district;
When b1. being decoded to little value district data, whenever decoding a little value district four-tuple, find out the corresponding watermark of code word of this four-tuple according to the table code word of the little value district Huffman shown in the table 1 and watermark mapping relations, be the watermark that extracts.If the corresponding code word of the four-tuple that decodes is the untapped code word in the table 1, then do not extract watermark;
C1. after all frames in the mp3 file all pass through the watermark extracting operational processes, all watermark informations that extract are merged into watermark file.
Further, described code word classification, the object of its classification is 2 the Huffman tables for the little value district data of encoding in the MP3 coding standard, criteria for classification is: the Hamming weight that calculates each four-tuple in each Huffman table, the corresponding code word of the four-tuple that Hamming weight is identical is classified as same class, if code word number is less than 4 in certain class, then all code words in such are included in other these classes.Classification results is as shown in table 2.
Further, the method for building up of little value district Huffman table code word and watermark mapping relations is as follows:
Simplify for the code word classification results in the little value district Huffman table, so that the code word number in every class is 2 n(n 〉=2).Namely on the classification results of table 2, the code word in every class is chosen suitable 2 nThen individual code word sets up the mapping relations of these code words and watermark.Code word need to satisfy when choosing in each classification: selected 2 nIndividual code word, its Hamming distance each other must be minimum, and is minimum to guarantee the modification to original audio behind the embed watermark.The mapping relations of code word and watermark are as shown in table 1 in the final little value district Huffman table of setting up.
Further, in the described watermark embed process, according to code word and watermark mapping relations embed watermark in the little value district Huffman table, its concrete meaning is as follows:
4 four-tuple of Hamming weight as 1 describe as example in the little value district Huffman Table A of table 1, and all classes among the table B of all classes of other in the Table A of table 1 and table 1 all can be by that analogy.Concrete steps are:
If a2. 2 bit watermark to be embedded are 00, no matter then what the corresponding code word of the current four-tuple that decodes is, it is revised as four-tuple 0001 corresponding code word 0101;
If b2. 2 bit watermark to be embedded are 01, no matter then what the corresponding code word of the current four-tuple that decodes is, it is revised as four-tuple 0010 corresponding code word 0100;
If c2. 2 bit watermark to be embedded are 10, no matter then what the corresponding code word of the current four-tuple that decodes is, it is revised as four-tuple 0100 corresponding code word 0110;
If d2. 2 bit watermark to be embedded are 11, no matter then what the corresponding code word of the current four-tuple that decodes is, it is revised as four-tuple 1000 corresponding code words 0111;
If the four-tuple that e2. decodes is among the Table A of table 1 and the table B during untapped code word, keep the corresponding code word of this four-tuple constant, do not carry out any operation.
Further, in the described watermark extraction process, extract watermark according to code word and watermark mapping relations in the little value district Huffman table, its concrete meaning is as follows:
4 four-tuple of Hamming weight as 1 describe as example in the little value district Huffman Table A of table 1, and all classes among all classes of other in the Table A and the table B all can be by that analogy.Concrete steps are:
If the four-tuple that a3. decodes is 0001, then extract 2 bit watermark 00;
If the four-tuple that b3. decodes is 0010, then extract 2 bit watermark 01;
If the four-tuple that c3. decodes is 0100, then extract 2 bit watermark 10;
If the four-tuple that d3. decodes is 1000, then extract 2 bit watermark 11;
If the four-tuple that e3. decodes is among the Table A of table 1 and the table B during untapped code word, there is not the watermark can be for extracting.
Further, set up in the process of code word and watermark mapping relations in the little value district Huffman table, choose 2 of Hamming distance minimum in every class nDuring individual code word, when run into other 2 nWhen the combination of individual code word has the characteristic of Hamming distance minimum equally, the combination in the table 2 can be chosen, also the code word combination that other satisfy this characteristic can be chosen.
The advantage that the present invention compared with prior art has is:
(1) a kind of MP3 water mark method for improving mp3 file watermark embedding capacity of the present invention, employed difference according to Hamming weight is carried out the method for little value district Huffman table code word classification, more can embody the information characteristic of code word in the little value district Huffman table, to after the mapping relations of MP3 watermark set up and watermarking algorithm has good booster action.
(2) a kind of MP3 water mark method for improving mp3 file watermark embedding capacity of the present invention, the little value district Huffman table code word of setting up and the mapping relations of watermark, comprise 4 code words in every class, corresponding 2 bit watermark, thereby, have greatly improved according to the capacity of this mapping relations embed watermark.
(3) a kind of MP3 water mark method for improving mp3 file watermark embedding capacity of the present invention, it is at little value district embed watermark, more traditional MP3 watermarking algorithm is less to the quality influence of MP3 audio frequency at large value district embed watermark, thereby the audio quality behind the embed watermark is better.
(4) a kind of MP3 water mark method for improving mp3 file watermark embedding capacity of the present invention, employed watermark embedding method in conjunction with the MP3 decoding process, the watermarking algorithm that carries out again the MP3 compression in conjunction with embed watermark in the watermark of MP3 compression encoding process and the original audio watermark, its real-time better effects if.
Description of drawings
Fig. 1 is watermark embedding of the present invention and watermark extracting process flow diagram.
Embodiment
The present invention is further described in detail below in conjunction with the drawings and specific embodiments.
The general frame figure of a kind of MP3 water mark method for improving mp3 file watermark embedding capacity of the present invention as shown in Figure 1.Overall flow mainly comprises watermark embedding and watermark extracting two large divisions.The method specifically comprises code word classification, code word and watermark mapping, watermark embed process and watermark extraction process; The code word classification is that all code words in the little value district Huffman table are divided according to the Hamming weight difference; Code word and watermark mapping is to set up mapping relations between code word and the watermark according to code word classification results and Hamming distance minimum principle; Watermark embed process is: before watermark embeds, code word in the little value district Huffman that uses in the MP3 cataloged procedure table is classified according to code word Hamming weight difference, then set up mapping relations between code word and watermark according to classification results and Hamming distance minimum principle; In telescopiny, successively the every frame data in the mp3 file are decoded, when being decoded to little value district, whenever decode a little value district quaternary class value, just according to the mapping relations embed watermark between code word and watermark, then amended code word is write mp3 file; At last, the MP3 frame that all is disposed reconstitutes the mp3 file that contains watermark; Watermark extraction process is: equally the every frame data in the mp3 file are decoded, when being decoded to little value district data, whenever decoding a little value district quaternary class value and just extract watermark according to the mapping relations between code word and watermark.
Wherein, the code word classification is that all code words in the little value district Huffman table are divided according to Hamming weight is different with Hamming distance, criteria for classification is: the Hamming weight that calculates each four-tuple in each Huffman table, the corresponding code word of the four-tuple that Hamming weight is identical is classified as same class, if code word number is less than 4 in certain class, then all code words in such are included into not in the use classes.Classification results is as shown in table 2.
Code word classification in little value district Huffman among the present invention table and code word and watermark mapping relations to set up process as follows:
Wherein, set up the mapping relations of little value district Huffman table code word and watermark, at first need the code word classification results in the little value district Huffman table is simplified, so that the code word number in every class is 2 n(n 〉=2).Namely on the classification results of table 2, the code word in every class is chosen suitable 2 nThen individual code word sets up the mapping relations of these code words and watermark.Code word need to satisfy when choosing in each classification: selected 2 nIndividual code word, its Hamming distance each other must be minimum, and is minimum to the modification of original audio to guarantee behind the embed watermark.The mapping relations of code word and watermark are as shown in table 1 in the final little value district Huffman table of setting up.In addition, choose 2 of Hamming distance minimum in every class nDuring individual code word, when run into other 2 nWhen the combination of individual code word has the characteristic of Hamming distance minimum equally, the combination in the table 2 can be chosen, also the code word combination that other satisfy this characteristic can be chosen.
The concrete steps of the watermark embed process among the present invention are as follows:
A. read in mp3 file, it is carried out the MP3 decoding operation.Carry out according to the decoding process in the MP3 standard in the decode procedure, each frame data in the mp3 file after the large value district data of having decoded, are carried out the decode operation in little value district;
When b. being decoded to little value district data, whenever decoding a little value district four-tuple, find out target code word behind the embed watermark according to the little value district Huffman shown in the table 1 table code word and watermark mapping relations, this target code word will be replaced the corresponding original code word of four-tuple.If the corresponding code word of the four-tuple that decodes is the untapped code word in the table 1, then keep its code word constant; Wherein, according to code word and watermark mapping relations embed watermark in the little value district Huffman table, its concrete meaning is as follows:
4 four-tuple of Hamming weight as 1 describe as example in the little value district Huffman Table A of table 1, and all classes among other classes in the Table A of table 1 and the table B of table 1 all can be by that analogy.Concrete steps are:
If a2. 2 bit watermark to be embedded are 00, no matter then what the corresponding code word of the current four-tuple that decodes is, it is revised as four-tuple 0001 corresponding code word 0101;
If b2. 2 bit watermark to be embedded are 01, no matter then what the corresponding code word of the current four-tuple that decodes is, it is revised as four-tuple 0010 corresponding code word 0100;
If c2. 2 bit watermark to be embedded are 10, no matter then what the corresponding code word of the current four-tuple that decodes is, it is revised as four-tuple 0100 corresponding code word 0110;
If d2. 2 bit watermark to be embedded are 11, no matter then what the corresponding code word of the current four-tuple that decodes is, it is revised as four-tuple 1000 corresponding code words 0111;
If the four-tuple that e2. decodes is among the Table A of table 1 and the table B during untapped code word, keep the corresponding code word of this four-tuple constant, do not carry out any operation.
C. after the little value district data in this frame being disposed according to step b, jointly reconstituting new frame data with other data of large value district's data and present frame, and jump to the processing that step b carries out next frame;
D. after all frames in the mp3 file all pass through the processing of watermark embedding operation, new MP3 frame is reconstituted the mp3 file that contains watermark.
The concrete steps of the watermark extraction process among the present invention are as follows:
A1. read in mp3 file, it is carried out the MP3 decoding operation.Carry out according to the decoding process in the MP3 standard in the decode procedure, each frame data in the mp3 file after large value district data are played in decoding, are carried out the decode operation in little value district;
When b1. being decoded to little value district data, whenever decoding a little value district four-tuple, find out the corresponding watermark of code word of this four-tuple according to the table code word of the little value district Huffman shown in the table 1 and watermark mapping relations, be the watermark that extracts.If the corresponding code word of the four-tuple that decodes is the untapped code word in the subordinate list 1, then do not extract watermark; Wherein, extract watermark according to code word and watermark mapping relations in the little value district Huffman table, its concrete meaning is as follows:
4 four-tuple of Hamming weight as 1 describe as example in the little value district Huffman Table A of table 1, and all classes among other classes in the Table A of table 1 and the table B of table 1 all can be by that analogy.Concrete steps are:
If the four-tuple that a3. decodes is 0001, then extract 2 bit watermark 00;
If the four-tuple that b3. decodes is 0010, then extract 2 bit watermark 01;
If the four-tuple that c3. decodes is 0100, then extract 2 bit watermark 10;
If the four-tuple that d3. decodes is 1000, then extract 2 bit watermark 11;
If the four-tuple that e3. decodes is among the Table A of table 1 and the table B during untapped code word, there is not the watermark can be for extracting.
C1. after all frames in the mp3 file all pass through the processing of watermark embedding operation, all watermark informations that extract are merged into watermark file.
In table 3, the algorithm that is used for two pieces of papers relatively and is the at present newer paper about the MP3 watermark: Yan2009 representative is that the people such as Yan has proposed a kind of improvement algorithm for the hidden writing tools of MP3Stego (Y.Diqun in 2009, W.Rangding, and Z.Liguang.Quantization step parity-based steganography for mp3 audio[J] Fundamenta Informaticae, 2009,97 (1): 1-14), eliminated potential endless loop in the MP3Stego algorithm; The algorithm of Yan2011 representative be Yan and Wang proposed in 2011 a kind of large value district Huffman table replace the MP3 watermark selected (D.Yan and R.Wang.Huffman table swapping-based steganograpy for mp3 audio[J] Multimedia Tools and Applications, 2011, pages 1-15.), its algorithm on capacity more in the past the MP3 watermarking algorithm improve a lot.Test employed testing audio and comprised the audio frequency of different-style in 5, be respectively: classical, blues, country, folk and pop.
Experimental result from table 3 can find out, the MP3 watermark based on code word in the little value district Huffman table and watermark mapping relations in this paper has very large raising than the algorithm of the people such as Yan proposition on embedding capacity.
Wherein, table 1 is code word and watermark mapping relations table in the little value district Huffman table of setting up among the present invention;
The code word classification results of table 2 among the present invention little value district Huffman being shown;
The experimental result signal table that table 3 is relatively gone up at maximum embedding capacity for the algorithm of the algorithm among the present invention and Yan and Wang.
Table 1
Figure BDA0000090126280000081
Table 2
Figure BDA0000090126280000082
Table 3
Figure BDA0000090126280000091
The part that the present invention does not elaborate belongs to techniques well known.

Claims (6)

1.一种用于提高MP3文件中水印嵌入容量的MP3水印方法，该方法基于小值区Huffman表码字与水印映射关系，其特征在于：该方法包括码字分类、码字与水印映射、水印嵌入过程和水印提取过程；其中，所述的码字分类是对小值区Huffman表中的所有码字按照汉明重量不同进行划分；所述的码字与水印映射是根据码字分类结果和汉明距离最小原则建立码字与水印之间的映射关系；所述的水印嵌入过程为：水印嵌入之前，对MP3编码过程中使用的小值区Huffman表中的码字按照码字汉明重量不同进行分类，然后按照分类结果与汉明距离最小原则建立码字和水印间的映射关系；在水印嵌入过程中，依次对MP3文件中的每帧数据进行解码，在解码到小值区时，每解码出一个小值区四元组值，就按照码字与水印间的映射关系嵌入水印，然后将修改后的码字写入MP3文件；最后，将所有处理完毕的MP3帧重构成含水印的MP3文件；所述的水印提取过程为：同样对MP3文件中的每帧数据进行解码，当解码到小值区数据时，每解码出一个小值区四元组值就按照码字与水印间的映射关系来提取水印；1. a kind of MP3 watermark method that is used to improve watermark embedding capacity in the MP3 file, this method is based on small-value area Huffman table codeword and watermark mapping relation, it is characterized in that: this method comprises codeword classification, codeword and watermark mapping, Watermark embedding process and watermark extraction process; wherein, the code word classification is to divide all code words in the small-value area Huffman table according to the different Hamming weights; the code word and watermark mapping is based on the code word classification result Establish the mapping relationship between the codeword and the watermark with the minimum principle of Hamming distance; the described watermark embedding process is: before the watermark is embedded, the codeword in the small-value area Huffman table used in the MP3 encoding process is according to the codeword Hamming Classify with different weights, and then establish the mapping relationship between codewords and watermarks according to the principle of the minimum distance between the classification results and Hamming distance; , each time a quadruple value in the small value area is decoded, the watermark is embedded according to the mapping relationship between the codeword and the watermark, and then the modified codeword is written into the MP3 file; finally, all processed MP3 frames are reconstructed into The MP3 file of watermark; The described watermark extracting process is: every frame data in the MP3 file is decoded equally, when decoding to small-value area data, whenever decoding a small-value area quadruple value just according to code word and The mapping relationship between watermarks is used to extract watermarks; 其中，所述的水印嵌入过程具体步骤如下：Wherein, the specific steps of the watermark embedding process are as follows: a.读入MP3文件，对其进行MP3解码操作；解码过程按照MP3标准中的解码流程进行，对MP3文件中的每一帧数据在解码完大值区数据后，进行小值区的解码操作；a. Read in the MP3 file and perform MP3 decoding operation on it; the decoding process is carried out according to the decoding process in the MP3 standard, and after decoding the data of the large value area for each frame of data in the MP3 file, perform the decoding operation of the small value area ; b.解码到小值区数据时，每当解码出一个小值区四元组，按照表1中所示的小值区Huffman表码字与水印映射关系找出嵌入水印后的目标码字，该目标码字将替换四元组所对应的原始码字；若解码出的四元组所对应的码字是表1中的未使用的码字，则保留其码字不变；b. When decoding the data in the small value area, whenever a small value area quaternion is decoded, find out the target code word after the watermark is embedded according to the mapping relationship between the small value area Huffman table code word and the watermark shown in Table 1, This target codeword will replace the original codeword corresponding to the quadruple; if the codeword corresponding to the decoded quadruple is an unused codeword in Table 1, then keep its codeword unchanged; c.将本帧中的小值区数据按照步骤b处理完毕后，与大值区数据以及当前帧的其他数据共同重构成新的帧数据，并跳到步骤b进行下一帧的处理；c. After processing the small-value area data in this frame according to step b, reconstruct new frame data together with the large-value area data and other data of the current frame, and skip to step b to process the next frame; d.当MP3文件中的所有帧都经过水印嵌入操作处理后，将新的MP3帧重构成含水印的MP3文件；d. After all the frames in the MP3 file are processed through the watermark embedding operation, the new MP3 frame is reconstructed into a watermarked MP3 file; 所述的水印提取过程具体步骤如下：The specific steps of the watermark extraction process are as follows: a1.读入MP3文件，对其进行MP3解码操作；解码过程中按照MP3标准中的解码流程进行，对MP3文件中的每一帧数据在解码完大值区数据后，进行小值区的解码操作；a1. Read in the MP3 file, and perform MP3 decoding operation on it; during the decoding process, follow the decoding process in the MP3 standard, and decode the small value area after decoding the data in the large value area for each frame of data in the MP3 file operate; b1.解码到小值区数据时，每当解码出一个小值区四元组，按照表1中所示的小值区Huffman表码字与水印映射关系找出该四元组的码字所对应的水印，即为提取出的水印；若解码出的四元组所对应的码字是表1中的未使用的码字，则不提取水印；b1. When decoding the data in the small-value area, whenever a small-value area quadruple is decoded, find out the code word of the quadruple according to the mapping relationship between the small-value area Huffman table code word and the watermark shown in Table 1 The corresponding watermark is the extracted watermark; if the codeword corresponding to the decoded quadruple is an unused codeword in Table 1, the watermark is not extracted; 表1Table 1
Figure FDA00001674145000021
Figure FDA00001674145000021
 c1.当MP3文件中的所有帧都经过水印提取操作处理后，将所有提取出的水印信息合并成水印文件。c1. After all the frames in the MP3 file are processed by the watermark extraction operation, all the extracted watermark information is combined into a watermark file. 2.根据权利要求1所述的一种用于提高MP3文件中水印嵌入容量的MP3水印方法，其特征在于：所述的码字分类中，分类的对象为MP3编码标准中的用于编码小值区数据的2个Huffman表，分类标准为：计算每个Huffman表中各四元组的汉明重量，将汉明重量相同的四元组所对应的码字归为同一类，若某类中码字个数小于4，则将该类中的所有码字归入其他这个类中；分类结果如表2所示。2. a kind of MP3 watermark method that is used to improve watermark embedding capacity in the MP3 file according to claim 1, is characterized in that: in the described code word classification, the object of classification is for encoding small The two Huffman tables of the value area data, the classification standard is: calculate the Hamming weight of each quadruple in each Huffman table, and classify the codewords corresponding to the quadruples with the same Hamming weight into the same category, if a certain category If the number of middle codewords is less than 4, then all codewords in this class are classified into other classes; the classification results are shown in Table 2. 表2Table 2
Figure FDA00001674145000031
Figure FDA00001674145000031
 3.根据权利要求1所述的一种用于提高MP3文件中水印嵌入容量的MP3水印方法，其特征在于：其中，小值区Huffman表码字与水印映射关系的建立方法如下：3. a kind of MP3 watermark method that is used to improve watermark embedding capacity in the MP3 file according to claim 1, is characterized in that: wherein, the establishment method of small-value area Huffman table code word and watermark mapping relationship is as follows: 对于小值区Huffman表中的码字分类结果进行简化，使得每类中的码字个数为2n，其中，n≥2；即在权利要求2的表2的分类结果上，对每类中的码字选取合适的2n个码字，然后建立这些码字与水印的映射关系；每个分类中码字选取时需要满足：所选取的2n个码字，其相互之间的汉明距离必须最小，以确保嵌入水印后对原始音频的修改最小；最终建立的小值区Huffman表中码字与水印的映射关系如权利要求1的表1所示。Simplify the codeword classification results in the small-value area Huffman table, so that the number of codewords in each class is 2 n , wherein, n≥2; that is, on the classification results of table 2 of claim 2, for each class Select the appropriate 2 n codewords in the codewords, and then establish the mapping relationship between these codewords and watermarks; the selection of codewords in each category needs to meet: the selected 2 n codewords, the mutual Chinese The distance must be the smallest to ensure that the modification of the original audio is the smallest after the watermark is embedded; the mapping relationship between the codeword and the watermark in the Huffman table of the small-value area finally established is as shown in Table 1 of claim 1. 4.根据权利要求1所述的一种用于提高MP3文件中水印嵌入容量的MP3水印方法，其特征在于：所述的水印嵌入过程中，根据小值区Huffman表中码字与水印映射关系嵌入水印，其具体含义如下：4. a kind of MP3 watermarking method that is used to improve watermark embedding capacity in the MP3 file according to claim 1, is characterized in that: in the described watermark embedding process, according to code word and watermark mapping relation in the small-value area Huffman table Embed watermark, its specific meaning is as follows: 以权利要求1的表1的小值区Huffman表A中汉明重量为1的4个四元组为例进行说明，权利要求1的表1的表A中的其他所有类和权利要求1的表1的表B中的所有类均可以此类推；具体步骤为：Take the four quadruples with a Hamming weight of 1 in the small value area Huffman Table A of Table 1 of Claim 1 as an example, all other classes in Table A of Table 1 of Claim 1 and the All classes in Table B of Table 1 can be deduced like this; the specific steps are: a2.若待嵌入的2比特水印为00，则不论当前解码出的四元组所对应的码字是什么，将其修改为四元组0001所对应的码字0101；a2. If the 2-bit watermark to be embedded is 00, no matter what the codeword corresponding to the currently decoded quadruple is, modify it to the codeword 0101 corresponding to the quadruple 0001; b2.若待嵌入的2比特水印为01，则不论当前解码出的四元组所对应的码字是什么，将其修改为四元组0010所对应的码字0100；b2. If the 2-bit watermark to be embedded is 01, no matter what the codeword corresponding to the currently decoded quadruple is, modify it to the codeword 0100 corresponding to the quadruple 0010; c2.若待嵌入的2比特水印为10，则不论当前解码出的四元组所对应的码字是什么，将其修改为四元组0100所对应的码字0110；c2. If the 2-bit watermark to be embedded is 10, no matter what the codeword corresponding to the currently decoded quadruple is, modify it to the codeword 0110 corresponding to the quadruple 0100; d2.若待嵌入的2比特水印为11，则不论当前解码出的四元组所对应的码字是什么，将其修改为四元组1000所对应的码字0111；d2. If the 2-bit watermark to be embedded is 11, no matter what the codeword corresponding to the currently decoded quadruple is, modify it to the codeword 0111 corresponding to the quadruple 1000; e2.若解码出的四元组是权利要求1的表1的表A和表B中未使用的码字时，保留该四元组所对应的码字不变，不进行任何操作。e2. if the decoded quadruple is an unused codeword in Table A and Table B of Table 1 of claim 1, keep the codeword corresponding to the quadruple unchanged, and do not perform any operation. 5.根据权利要求1所述的一种用于提高MP3文件中水印嵌入容量的MP3水印方法，其特征在于：所述的水印提取过程中，根据小值区Huffman表中码字与水印映射关系提取水印，其具体含义如下：5. a kind of MP3 watermark method that is used to improve watermark embedding capacity in the MP3 file according to claim 1, is characterized in that: in the described watermark extracting process, according to code word and watermark mapping relation in the Huffman table of small value area Extract the watermark, its specific meaning is as follows: 以权利要求1的表1的小值区Huffman表A中汉明重量为1的4个四元组为例进行说明，权利要求1的表1的表A中的其他所有类和权利要求1的表1的表B中的所有类均可以此类推；具体步骤为：Take the four quadruples with a Hamming weight of 1 in the small value area Huffman table A of Table 1 of claim 1 as an example, all other classes in Table A of Table 1 of claim 1 and the All classes in Table B of Table 1 can be deduced like this; the specific steps are: a3.若解码出的四元组为0001，则提取2比特水印00；a3. If the decoded quadruple is 0001, extract the 2-bit watermark 00; b3.若解码出的四元组为0010，则提取2比特水印01；b3. If the decoded quadruple is 0010, extract the 2-bit watermark 01; c3.若解码出的四元组为0100，则提取2比特水印10；c3. If the decoded quadruple is 0100, then extract the 2-bit watermark 10; d3.若解码出的四元组为1000，则提取2比特水印11；d3. If the decoded quadruple is 1000, extract the 2-bit watermark 11; e3.若解码出的四元组是权利要求1的表1的表A和表B中未使用的码字时，没有水印可供提取。e3. If the decoded quaternion is an unused codeword in Table A and Table B of Table 1 of claim 1, there is no watermark for extraction. 6.根据权利要求3所述的一种用于提高MP3文件中水印嵌入容量的MP3水印方法，其特征在于：建立小值区Huffman表中码字与水印映射关系的过程中，每类中选取汉明距离最小的2n个码字时，当遇到其他2n个码字的组合同样具有汉明距离最小的特性时，可选取表2中的组合，也可选取其他满足该特性的码字组合。6. a kind of MP3 watermarking method that is used to improve watermark embedding capacity in the MP3 file according to claim 3 is characterized in that: in the process of setting up the codeword and watermark mapping relationship in the small-value area Huffman table, select in every class When there are 2 n codewords with the smallest Hamming distance, when encountering the combination of other 2 n codewords that also has the characteristic of the smallest Hamming distance, you can choose the combination in Table 2, or choose other codes that satisfy this characteristic word combination.
CN2011102660642A 2011-09-09 2011-09-09 MP3 (Moving Picture Experts Group Audio Layer 3) watermarking method for improving watermark-embedding capacity in MP3 file Expired - Fee Related CN102314881B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2011102660642A CN102314881B (en) 2011-09-09 2011-09-09 MP3 (Moving Picture Experts Group Audio Layer 3) watermarking method for improving watermark-embedding capacity in MP3 file

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2011102660642A CN102314881B (en) 2011-09-09 2011-09-09 MP3 (Moving Picture Experts Group Audio Layer 3) watermarking method for improving watermark-embedding capacity in MP3 file

Publications (2)

Publication Number Publication Date
CN102314881A CN102314881A (en) 2012-01-11
CN102314881B true CN102314881B (en) 2013-01-02

Family

ID=45427991

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2011102660642A Expired - Fee Related CN102314881B (en) 2011-09-09 2011-09-09 MP3 (Moving Picture Experts Group Audio Layer 3) watermarking method for improving watermark-embedding capacity in MP3 file

Country Status (1)

Country Link
CN (1) CN102314881B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB2524784B (en) * 2014-04-02 2018-01-03 Law Malcolm Transparent lossless audio watermarking
CN110428846A (en) * 2019-07-08 2019-11-08 清华大学 Voice-over-net stream steganalysis method and device based on bidirectional circulating neural network

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1465157A1 (en) * 2003-03-31 2004-10-06 Sony United Kingdom Limited Digital audio watermark
CN101290773A (en) * 2008-06-13 2008-10-22 清华大学 Adaptive MP3 digital watermark embedding method
CN101894555A (en) * 2010-04-09 2010-11-24 中山大学 Watermark protection method for MP3 file

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20060023976A (en) * 2003-05-28 2006-03-15 코닌클리케 필립스 일렉트로닉스 엔.브이. Bit-stream watermarking
US8401972B2 (en) * 2006-10-27 2013-03-19 At&T Intellectual Property Ii, L.P. System and method of generating a per sale watermarking on digital files
US8259938B2 (en) * 2008-06-24 2012-09-04 Verance Corporation Efficient and secure forensic marking in compressed

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1465157A1 (en) * 2003-03-31 2004-10-06 Sony United Kingdom Limited Digital audio watermark
CN101290773A (en) * 2008-06-13 2008-10-22 清华大学 Adaptive MP3 digital watermark embedding method
CN101894555A (en) * 2010-04-09 2010-11-24 中山大学 Watermark protection method for MP3 file

Also Published As

Publication number Publication date
CN102314881A (en) 2012-01-11

Similar Documents

Publication Publication Date Title
US20040059918A1 (en) Method and system of digital watermarking for compressed audio
CN101206861B (en) Method for imbedding digital music production authentication information and method for authentication of said production
TW200935403A (en) Technique for encoding/decoding of codebook indices for quantized MDCT spectrum in scalable speech and audio codecs
CN103106901B (en) Audio digital steganography and extraction method in compressed domain based on index values
CN101211562B (en) Digital music works damage-free digital watermarking embedding and extraction method
Korycki Authenticity examination of compressed audio recordings using detection of multiple compression and encoders’ identification
CN100546198C (en) A Method for Embedding Authentication Information of Digital Music Works
Hu et al. Effective blind speech watermarking via adaptive mean modulation and package synchronization in DWT domain
CN104700840B (en) Watermark insertion in frequency domain for audio encoding/decoding/transcoding
Bibhu et al. Secret key watermarking in WAV audio file in perceptual domain
Yan et al. Huffman table swapping-based steganograpy for MP3 audio
CN102314881B (en) MP3 (Moving Picture Experts Group Audio Layer 3) watermarking method for improving watermark-embedding capacity in MP3 file
CN101894555A (en) Watermark protection method for MP3 file
CN103745725B (en) A kind of audio frequency watermark embedding grammar based on constant Q transform
Luo et al. A robust watermarking method for MPEG-4 SLS audio
Huang et al. AAC audio compression detection based on QMDCT coefficient
Wei et al. Controlling bitrate steganography on AAC audio
CN104485109B (en) A kind of compressed domain audio digital extraction method based on index value
Quan et al. Data hiding in MPEG compressed audio using wet paper codes
Tsai et al. An effective watermarking method based on energy averaging in audio signals
Yang et al. Lossless and secure watermarking scheme in MP3 audio by modifying redundant bit in the frames
Kirbiz et al. Decode-time forensic watermarking of AAC bitstreams
Liu et al. A robust recoverable algorithm used for digital speech forensics based on DCT
CN1987998A (en) A Cepstrum Sound Watermark Embedding and Extraction Method for Protecting Sound Copyright
Ho et al. An audio watermarking algorithm based on significant component modulation

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20130102

Termination date: 20150909

EXPY Termination of patent right or utility model