CN104331448A - Encrypted domain image retrieval method based on JPEG (Joint Picture Expert Group) image alternating current coefficient histograms - Google Patents

Encrypted domain image retrieval method based on JPEG (Joint Picture Expert Group) image alternating current coefficient histograms Download PDF

Info

Publication number
CN104331448A
CN104331448A CN201410591470.XA CN201410591470A CN104331448A CN 104331448 A CN104331448 A CN 104331448A CN 201410591470 A CN201410591470 A CN 201410591470A CN 104331448 A CN104331448 A CN 104331448A
Authority
CN
China
Prior art keywords
image
bit stream
coefficient
component
represent
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.)
Granted
Application number
CN201410591470.XA
Other languages
Chinese (zh)
Other versions
CN104331448B (en
Inventor
程航
张新鹏
童人婷
徐淼
王子驰
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.)
University of Shanghai for Science and Technology
Original Assignee
University of Shanghai for Science and Technology
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 University of Shanghai for Science and Technology filed Critical University of Shanghai for Science and Technology
Priority to CN201410591470.XA priority Critical patent/CN104331448B/en
Publication of CN104331448A publication Critical patent/CN104331448A/en
Application granted granted Critical
Publication of CN104331448B publication Critical patent/CN104331448B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/58Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/583Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T9/00Image coding

Abstract

The invention discloses an encrypted domain image retrieval method based on JPEG (Joint Picture Expert Group) image alternating current coefficient histograms. The method includes the process of encryption done by an image content owner, the process of image query done by an authorized user and the process of encrypted image retrieval done by a retrieval service provider, and Fig 1 is the general flow block diagram of the method. The image content owner encrypts the DCT coefficient bit stream of a JPEG image, reconstructs the bit stream to generate an encrypted file, and then submits the encrypted file to the retrieval service provider; the authorized user submits an encrypted image to be queried for retrieval, and decrypts and recovers the returned encrypted image; by calculating the distances between the alternating current coefficient histograms of the image to be queried and all the images in an encrypted image database one by one, the retrieval service provider returns an image most approximate to the histogram of the image to be queried to the authorized user. While protecting the privacy of image data, the method realizes encrypted image retrieval; moreover, computation ad communication overheads and file storage spaces are reduced, and meanwhile, the security of the color information of encrypted images is taken into consideration.

Description

Based on the histogrammic ciphertext domain image search method of jpeg image ac coefficient
Technical field
The present invention relates to a kind of based on the histogrammic ciphertext domain image search method of jpeg image ac coefficient, utilize digital signal processing and encryption technology adjustment jpeg image bitstream structure to carry out image retrieval in ciphertext domain.
Background technology
Along with the arrival in cloud computing epoch, user can carry out store and management the service provider that complicated for this locality data management system (such as: the management system such as Email, individual medical records, financial transaction) is contracted out to high in the clouds whenever and wherever possible, but data are stored in and where, how process, user is unwitting, cause user thus to the worry of data-privacy, certainly will require first to carry out relevant treatment to submitting to service provider after data encryption again, in case private data is revealed.Thisly above protected data privacy should ensure that the demand of data normal process is for being a great challenge based on conventional IR service expressly again.In cipher-text information searching system, text data after encryption or multi-medium data form cryptograph files database purchase in server, retrieval user submits ciphertext key word, ciphertext feature or encrypt data to server, and server returns the cryptograph files meeting retrieval and require.Searching ciphertext can realize the storage and retrieval of file under the condition not revealing cleartext information by external storage resources and computational resource, be the important foundation that secure cloud stores.In this relatively emerging research field of ciphertext domain information retrieval, emerge a large amount of achievements at present, mainly contain two-part work: the key word file retrieval of ciphertext domain and the image retrieval of ciphertext domain.The information retrieval major part work of ciphertext domain concentrates on the former, and compares less about the image retrieval technologies research report of ciphertext domain.
The image search method of existing ciphertext domain mainly contains three kinds, the method of the first ciphertext image retrieval is: utilize the word bag model of text retrieval technique to represent that characteristics of image describes, associating order-preserving encryption and min-hash function realize Security Index, not only protect lexical feature vector privacy but also support that result for retrieval sorts; The method of the second ciphertext image retrieval is: carry out the process of randomization as: position layer randomization, accidental projection or a random primitive encoding to primitive image features, make apart from close feature still close together after randomization, and the image retrieval under utilizing this character to realize not revealing content conditions; The method of the third ciphertext image retrieval is: utilize the coefficient histogram unchangeability of discrete cosine transform same frequency before and after image encryption to realize the image retrieval of ciphertext domain.Can consult as Publication about Document:
[1] W. Lu, A. Swaminathan, A. L. Varna, and M. Wu, Enabling search over encrypted multimedia databases, In Proceedings of SPIE,vol. 7254, Conf. on Media Forensics and Security, January 2009.
[2] W. Lu, A. L. Varna, A. Swaminathan, and M. Wu, Secure image retrieval through feature protection, IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2009,1533-1536, 2009.
[3] X. Zhang, H. Cheng, Histogram-based Retrieval for Encrypted JPEG Images, Signal and Information Processing (ChinaSIP), 2014 IEEE China Summit & International Conference on. IEEE, 2014.
But in above-mentioned first and second kind of method, image characteristics extraction, feature encryption, image encryption will separately process separately, and the characteristic sum image after encryption will be stored in server simultaneously, can increase computation complexity and communication and storage overhead like this; In the third method above-mentioned, although characteristics of image does not need additional extractions and encryption, jpeg image same frequency coefficient value scrambling encryption has upset the continuity of the JPEG compressed encoding of original image, causes jpeg image file to increase and colouring information leakage.
Summary of the invention
The object of the invention is to overcome the deficiencies in the prior art, provide a kind of based on the histogrammic ciphertext domain image search method of jpeg image ac coefficient, the method is based on the bitstream structure of jpeg image compressed encoding, in conjunction with stream cipher XOR encryption method and scrambling encryption method, obtain ciphertext jpeg image.Utilize jpeg image encrypt before and after ac coefficient histogram unchangeability, realize ciphertext image retrieval, do not increasing ciphertext image file size and take into account simultaneously the colouring information safety of ciphertext graph picture.
In order to achieve the above object, design of the present invention is: the present invention is divided into three parts, and one is picture material owner ciphering process; Two is authorized user query image processes; Three is that retrieval service supplier retrieves ciphertext image process:
The picture format that the present invention adopts is jpeg format, and JPEG is the english abbreviation of JPEG (joint photographic experts group) (Joint Picture Expert Group), is a kind of compression and coding standard of still image.Due to jpeg image high compression ratio, make it be used widely in multimedia and WWW, therefore have important practical significance and using value at the enterprising line retrieval of the image of jpeg format.
According to foregoing invention design, the present invention adopts following technical proposals:
A kind of based on the histogrammic ciphertext domain image search method of jpeg image ac coefficient, it is characterized in that specific implementation step is as follows:
(1), picture material owner ciphering process.Read the JPEG coding stream of the expressly arbitrary image of image data base, obtain the DC coefficient bit stream after its discrete cosine transform and ac coefficient bit stream two parts, then stream cipher XOR encryption method and scrambling encryption method is adopted to be encrypted DC coefficient bit stream and ac coefficient bit stream respectively, and the jpeg image after generating encryption is reconfigured to the bit stream after encryption, finally the ciphertext image data base after all images encryption is submitted to retrieval service supplier;
(2), authorized user query image process.First user needs to obtain picture material owner mandate before inquiry expressly image, become authorized user, then the image that will inquire about is submitted to the picture material owner to encrypt accordingly, after corresponding ciphertext graph picture to be obtained, just can directly get on to retrieve matching image at image retrieval server, finally the plaintext image close with query image be obtained to the ciphertext result deciphering returned;
(3), retrieval service supplier retrieves ciphertext image process.The ciphertext image data base that the retrieval service supplier store and management picture material owner submits to, and provide ciphertext graph as retrieval service.Retrieval service supplier extracts ac coefficient bit stream from the cryptogram search image file bit stream that authorized user is submitted to, the corresponding ac coefficient histogram of cryptogram search image is obtained by JPEG decoding and mathematical statistics means, then cryptogram search image communication coefficient histogram is carried out corresponding one by one comparison with the known ac coefficient histogram of images all in ciphertext database, finally return the image the most close with image histogram to be checked to authorized user.
The present invention compared with prior art; there is following apparent outstanding substantive distinguishing features and remarkable advantage: the method utilizes the bitstream structure of jpeg image compressed encoding; have employed the particular encryption technology of jpeg image ac coefficient histogram unchangeability; not only achieve ciphertext image retrieval; protect the data-privacy of query image and image data base; and characteristics of image does not need to extract and encrypt these additional steps in advance; do not increasing file size, reduction computation complexity and communication overhead take into account the colouring information safety of ciphertext graph picture simultaneously.
Accompanying drawing explanation
Fig. 1 is the FB(flow block) based on the histogrammic ciphertext domain image search method of jpeg image ac coefficient of the present invention;
Fig. 2 is that parts of images sample in the plaintext image data base described in step (1-1) is (from Corel standard testing image database, size 256 × 384 or 384 × 256, comprise ten class images, as African, seabeach, building, motorbus, dinosaur, elephant, flower, horse, mountain, food, totally 1000 sub-pictures);
Fig. 3 is the sample format of jpeg image Y, U, V tri-components described in step (1-1): YUV411;
Fig. 4 is plaintext image to be encrypted in the image data base described in step (1-1) ;
Fig. 5 is the jpeg image described in step (1-2) dCcoefficient and aCthe coding structure of coefficient;
Fig. 6 is the VLI code table described in step (1-2);
Fig. 7 is the ciphertext graph picture described in step (1-4) ;
Fig. 8 is the user described in step (2-1) the image that need inquire about ;
Fig. 9 is the ciphertext graph picture described in step (2-1) ;
Figure 10 is the plaintext image in embodiment;
Figure 11 is the ciphertext graph picture that in embodiment, Figure 10 is corresponding.
Embodiment
Below in conjunction with accompanying drawing, the preferred embodiments of the present invention are elaborated.
Embodiment one:
See Fig. 1, this is based on the histogrammic ciphertext domain image search method of jpeg image ac coefficient, and its specific implementation step is as follows:
(1), picture material owner ciphering process: the JPEG coding stream reading the expressly arbitrary image of image data base, obtain the DC coefficient bit stream after its discrete cosine transform and ac coefficient bit stream two parts, then stream cipher XOR encryption method and scrambling encryption method is adopted to be encrypted DC coefficient bit stream and ac coefficient bit stream respectively, and the jpeg image after generating encryption is reconfigured to the bit stream after encryption, finally the ciphertext image data base after all images encryption is submitted to retrieval service supplier;
(2), authorized user query image process: first user needs to obtain picture material owner mandate before inquiry expressly image, become authorized user, then the image that will inquire about is submitted to the picture material owner to encrypt accordingly, after corresponding ciphertext graph picture to be obtained, just can directly get on to retrieve matching image at image retrieval server, finally the plaintext image close with query image be obtained to the ciphertext result deciphering returned;
(3), retrieval service supplier retrieves ciphertext image process: the ciphertext image data base that the retrieval service supplier store and management picture material owner submits to, and provides ciphertext graph as retrieval service; Retrieval service supplier extracts ac coefficient bit stream from the cryptogram search image file bit stream that authorized user is submitted to, the corresponding ac coefficient histogram of cryptogram search image is obtained by JPEG decoding and mathematical statistics means, then cryptogram search image communication coefficient histogram is carried out corresponding one by one comparison with the known ac coefficient histogram of images all in ciphertext database, finally return the image the most close with image histogram to be checked to authorized user.
Embodiment two:
The present embodiment is substantially identical with embodiment one, is to describe an example by reference to the accompanying drawings.Special feature is as follows:
The concrete operation step of described step (1) picture material owner ciphering process is as follows:
(1-1), read arbitrary jpeg image file in plaintext image data base, obtain the bit stream sequence corresponding to image, the plaintext image data base of employing is Corel standard image data storehouse, and its parts of images as shown in Figure 2; According to JPEG compression standard, one pair first will be transformed into other color spaces without the RGB cromogram of any process before compressed encoding, what jpeg image adopted is YUV color space, and the sample format of YUV is YUV411, as shown in Figure 3, i.e. the data sampling ratio of Y, U, V tri-components: 4 corresponding 1 U and 1 V of Y; Then, Y, U, V tri-components are carried out the fritter being divided into nonoverlapping 8 X 8 respectively, dct transform is carried out to each fritter, produce 64 coefficients: 1 dCcoefficient and 63 aCcoefficient, and right dCcoefficient and aCthe different quantization table of coefficient quantizes; To quantize after dCcoefficient carries out Differential pulse code modulation coding and entropy code, and after quantizing aCcoefficient then adopts RLC to encode and entropy code; Finally in conjunction with JPEG specify mark by certain form the bit stream produced after picture headers information and entropy code by byte-sized writing in files;
Note: RGB represents three kinds of color: R and represents red, and G represents green, and B represents blue; Y represents luminance component, U and V all represents color difference components, and dct transform represents discrete cosine transform, dCcoefficient represents DC coefficient, aCcoefficient represents ac coefficient, and RLC represents distance of swimming compressed encoding;
Suppose the plaintext image read as shown in Figure 4, the JPEG coding stream sequence of acquisition mainly comprises two-part information: File header information and image compression data , the wherein supplementary needed for JPEG encoding and decoding such as the main include file size of File header information, quantization table, Huffman code tables, and image compression data is original image data by data after compressed encoding, it is expressed as:
(1)
Wherein, represent image ? (sample format of YUV is 4:1:1 to block, for Y-component block is then made up of 48 X 8 pieces, U and V component is made up of 18 X 8 pieces) through entropy code compression after data structure, ( ) represent that 16 X 16 pieces is at image the position of middle horizontal ordinate, represent image height, ( ) represent that 16 X 16 pieces is at image the position of middle ordinate, represent image wide, represent the bit stream after block Y-component entropy code, block size 8 X 8, , represent the position that block puts in order, represent the bit stream after U and V component entropy code respectively;
(1-2), basis data structure, extract image ? block Y's dCcoefficient bit stream and aCcoefficient bit stream; If image ? block Y-component dCbit stream is , it is expressed as:
(2)
Wherein, represent image the in block after block Y-component entropy code dCcoefficient bit stream, , represent the position that block puts in order;
If image ? block Y-component aCbit stream is , it is expressed as:
(3)
Wherein, ( ) represent image the in block after block Y-component entropy code aCcoefficient bit stream, represent the position that block puts in order;
From formula (2) (3), ( ) by with form, it is expressed as:
(4)
When time fixing, be actually the bit stream after 64 DCT coefficient entropy codings corresponding to 8 X 8 size block, wherein contain 1 dCcoefficient and 63 aCcoefficient, dCcoefficient and aCcoefficient coding structure is identical, all has two parts to form: Huffman code stream and VLI code stream, as shown in Figure 5;
Wherein, the Huffman encoding of Huffman representation, be a kind of harmless elongated compaction coding method, VLI is variable length integer coding;
When time, bitstream structure expression formula is:
(5)
Wherein, represent image ? block in component dCcoefficient Huffman code stream, represent image ? block in component dCcoefficient VLI code stream, utilizes image y-component in header file information dCcoefficient Huffman code table, the readable taking-up of the picture material owner bit stream, then right bit stream is decoded, and obtaining should dCcoefficient is group number in VLI code table, and its numerical values recited equals actual storage binary value bit stream length, reads out bit stream, with reference to figure 6, obtains dCthe numerical value of coefficient before entropy code;
When time, bitstream structure expression formula is:
(6)
Wherein, represent in aCcoefficient individual coding unit corresponding Huffman code stream, represent the number of appearance zero continuously in Run-Length Coding, represent continuous zero rear first non-zero values corresponding group number in VLI code table, represent in aCcoefficient individual coding unit in corresponding VLI code stream, represent in aCthe number of coefficient Run-Length Coding unit, ; Utilize image y-component in header file information aCcoefficient Huffman code table, the readable taking-up of the picture material owner bit stream, then right bit stream is decoded, and obtaining should aCcoefficient in VLI code table group number (its value equals bit stream length, with reference to figure 6), read out bit stream;
By that analogy, read out further , can image be obtained ? block Y-component dCbit stream with aCbit stream ;
(1-3), according to (1-2) step, the picture material owner is from image readable taking-up the bit stream of U, V component corresponding to block , different unlike the Huffman code table of U, V component Huffman code table used and Y-component from the bit stream reading Y-component;
(1-4), based on (1-2) and (1-3) step, the picture material owner can obtain whole sub-picture middle Y, U, V component dCbit stream and aCbit stream; For image 's dCbit stream and aCbit stream, takes different encryption methods, can obtain ciphertext graph picture , as shown in Figure 7, specific as follows:
The picture material owner is by whole sub-picture middle Y, U, V component dCbit stream is combined into a new bit stream sequence , its expression formula is as follows:
(7)
Wherein, , , represent image respectively middle Y, U, V component dCbit stream sequence, ( ) represent bit stream the binary value of position;
If image corresponding bit stream length is , the method pair that the picture material owner adopts stream cipher XOR to encrypt be encrypted, namely according to encryption key producing a string length is pseudo-random binary bit stream with carry out xor operation, generating length is encryption after bit stream , xor operation meets following formula:
(8)
Namely , obtain ciphertext bit stream ;
For image middle Y, U, V component aCbit stream takes different encryption technologies and scrambling encryption method, extracts image according to formula (6) block bitstream structure in all 8 X 8 pieces of Y, U, V components aCcoefficient block bit stream, and respectively block bit stream Connecting groups is synthesized new bit stream sequence, be set to , , , expression formula is as follows respectively:
(9)
(10)
(11)
Wherein, represent image Y-component the block, block size 8 X the 8,1st aCcode stream section corresponding to coefficient coding unit, with reference to (1-2) step, represent the 1st aCbit stream part after coefficient coding unit Huffman encodes, represent expression the 1st aCbit stream part after coefficient coding unit VLI encodes, ( ) represent the position of 8 X 8 pieces horizontal ordinate in the y component, ( ) represent the position of 8 X 8 pieces ordinate in the y component;
represent that image U divides flow control block (block size 8 X 8) the 1st aCcode stream section corresponding to coefficient coding unit, with reference to (1-2) step, represent the 1st aCbit stream part after coefficient coding unit Huffman encodes, represent expression the 1st aCbit stream part after coefficient coding unit VLI encodes, ( ) represent the position of 8 X, 8 pieces of horizontal ordinates in U component, ( ) represent the position of 8 X, 8 pieces of ordinates in U component; Symbol implication in formula (11) is similar with formula (10);
Separately, image the size of corresponding Y, U, V component is respectively , ;
With ( ) code stream section integrally unit, the picture material owner uses encryption key right middle comprised code stream section carries out scrambling encryption, namely upsets code stream section exists position in bit stream, Y-component after note encryption aCbit stream is ; To U, V component aCbit stream , take identical encryption method to be encrypted, obtaining ciphertext bit stream is , , key used is respectively , ;
(1-5), according to the format standard of JPEG, by image header information and encryption after Y, U, V component bit stream carry out reconnecting combination, can image be obtained corresponding encrypted image bit stream sequence , its expression formula is:
(12)
Wherein, represent image header information, have neither part nor lot in encryption, its bit stream remains unchanged before encryption afterwards. represent length information, comprise bit stream , , , length and JPEG cataloged procedure in the figure place of filler, value exists with the form of bit stream;
Separately, mark in jpeg format and start with hexadecimal FF, after with the byte of a non-zero, when reading jpeg file in (1-1) step, if FF byte 0 below, then deletion value is this byte of 0, in this operation obtain encrypted image bit stream compares original image bit stream length reduce to a certain extent, ensure that the file size after encryption does not increase, even file size most image reduced on the contrary;
(1-6) other image encryption process, in image data base all and image cryptographic operation the same, thus obtain encryption after ciphertext image data base, finally ciphertext image data base is submitted to retrieval service supplier and carries out store and management.
The concrete operation step of described step (2) authorized user query image process is as follows:
(2-1), according to both sides' rendezvous protocol, need the user of inquiry service to obtain picture material supplier and authorize, become authorized user ;
(2-2), user the image of inquiry will be needed submit to image contents service person to be encrypted, the picture material owner adopts step (1-1), (1-2), (1-3), (1-4), (1-5) by user the plaintext image submitted to be encrypted to ciphertext graph picture , finally ciphertext graph picture with corresponding encryption key , , , return to user ;
(2-3), user by ciphertext graph picture submit to retrieval service supplier to retrieve, and the number that need return the most close image of announcement, wait for result for retrieval simultaneously;
(2-4), user the ciphertext graph picture that deciphering retrieval service supplier returns, if the ciphertext graph picture returned is (corresponding plaintext image is , see Fig. 4), user read ciphertext graph picture , obtain image according to step (1-5) header information and compressed data information, reference formula (12).Image is obtained according to jpeg marker header information , and do not do extra process.From image is obtained in bit stream in , , , the length of bit stream, is partitioned into according to respective bit stream length , , , four section streams.Reference formula (8), utilizes well-known key produce pseudo-random binary bit stream with carry out xor operation to decrypt , for , utilize key scramble the position of middle code stream section, recovers code stream section original position, namely decrypt bit stream.By that analogy, key is utilized , decrypt , bit stream.Decompose according to step (1-4) , , , , obtain image middle Y, U, V component 8 X 8 pieces is corresponding dCwith aCbit stream, utilizes formula (1) to be combined into image the packed data of entropy code , combining image header information namely image is obtained bit stream , its expression formula is:
(13)
Separately, according to JPEG coding standard, image bit stream is by byte writing in files, if occur, byte value is FF, and is not JPEG regulation mark, then need fill a value after FF byte is the byte of 0, namely fill in step (1-5) deleted fall value be the byte of 0, thus decrypt ciphertext graph picture original image if return how secondary ciphertext graph picture, be all decrypted operation recovery according to step (2-4) and go out corresponding original plaintext image;
The concrete operation step that described step (3) retrieval service supplier retrieves ciphertext image process is as follows:
(3-1), retrieval service supplier read authorized user the ciphertext graph picture to be checked submitted to (corresponding plaintext image is ), obtain in ciphertext graph picture in , , three section streams, acquisition process refer step (2-4), in conjunction with Huffman code table and VLI code table, extracts ciphertext graph picture y, U, V tri-components all aCcoefficient, and utilize mathematical statistics means to obtain image y, U, V tri-components aCthe histogram that coefficient is corresponding , , ;
Separately, adopt to use the same method and ask all images in ciphertext database corresponding aCcoefficient histogram;
(3-2), ciphertext graph picture is calculated one by one and ciphertext graph is as the Histogram distance between all images of lane database, this distance is the measure of two width image similarity, and between distance less image, similarity is higher, otherwise then similarity is lower.The present invention adopts ratio of precision euclidean distance is higher relative distance, i.e. city block distance, its expression formula is as follows:
(14)
Wherein, for image to be checked, for image arbitrary in image data base, for query image middle Y-component aCcoefficient is frequency, for query image middle Y-component aCcoefficient is frequency.One secondary jpeg image is made up of Y, U, V tri-components, utilizes formula (14) can try to achieve U, V two components respectively aCthe distance of coefficient histogram , , cumulative three component distances are as dimensioned plan picture and image the final distance of similarity , expression formula is as follows:
(15)
Wherein, respectively , , weight, take exhaustive method to make , , weight optimum in formula (15);
Formula (14) (15) are utilized to obtain ciphertext graph picture one by one and ciphertext graph is as Histogram distance (comprising Y, U, V component) between lane database all images;
(3-3), to the ciphertext graph picture required by (3-2) step and ciphertext graph sorts from small to large as the Histogram distance between all images of lane database, forward preferentially the returning of sequence number, specifically returns ciphertext graph picture in how many auxiliary image data storehouse by authorized user tell retrieval service supplier in advance.
In order to verify effect of the present invention, adopt of the present invention based on the histogrammic ciphertext domain image search method of jpeg image ac coefficient, ciphertext graph picture (Fig. 9 with Figure 11) has no to associate with corresponding expressly image (Fig. 8 with Figure 10) on colouring information, takes into account the visual security of image while not revealing picture material.At image bitstream encryption regrouping process, the deletion of 0 value byte (previous byte value is FF) ensure that the file size of ciphertext does not increase (even reducing).In addition, image characteristics extraction and encrypting step, without the need to prior independent operation, also greatly reduce computation complexity, reduce communication overhead and storage space.

Claims (4)

1., based on the histogrammic ciphertext domain image search method of jpeg image ac coefficient, its specific implementation step is as follows:
(1), picture material owner ciphering process: the JPEG coding stream reading the expressly arbitrary image of image data base, obtain the DC coefficient bit stream after its discrete cosine transform and ac coefficient bit stream two parts, then stream cipher XOR encryption method and scrambling encryption method is adopted to be encrypted DC coefficient bit stream and ac coefficient bit stream respectively, and the jpeg image after generating encryption is reconfigured to the bit stream after encryption, finally the ciphertext image data base after all images encryption is submitted to retrieval service supplier;
(2), authorized user query image process: first user needs to obtain picture material owner mandate before inquiry expressly image, become authorized user, then the image that will inquire about is submitted to the picture material owner to encrypt accordingly, after corresponding ciphertext graph picture to be obtained, just can directly get on to retrieve matching image at image retrieval server, finally the plaintext image close with query image be obtained to the ciphertext result deciphering returned;
(3), retrieval service supplier retrieves ciphertext image process: the ciphertext image data base that the retrieval service supplier store and management picture material owner submits to, and provides ciphertext graph as retrieval service; Retrieval service supplier extracts ac coefficient bit stream from the cryptogram search image file bit stream that authorized user is submitted to, the corresponding ac coefficient histogram of cryptogram search image is obtained by JPEG decoding and mathematical statistics means, then cryptogram search image communication coefficient histogram is carried out corresponding one by one comparison with the known ac coefficient histogram of images all in ciphertext database, finally return the image the most close with image histogram to be checked to authorized user.
2. according to claims 1 describe based on the histogrammic ciphertext domain image search method of jpeg image ac coefficient, it is characterized in that: the concrete operation step of described step (1) picture material owner ciphering process is as follows:
(1-1), read arbitrary jpeg image file in plaintext image data base, obtain the bit stream sequence corresponding to image, the plaintext image data base of employing is Corel standard image data storehouse; According to JPEG compression standard, one pair first will be transformed into other color spaces without the RGB cromogram of any process before compressed encoding, what jpeg image adopted is YUV color space, and the sample format of YUV is YUV411, i.e. the data sampling ratio of Y, U, V tri-components: 4 corresponding 1 U and 1 V of Y; Then, Y, U, V tri-components are carried out the fritter being divided into nonoverlapping 8 X 8 respectively, dct transform is carried out to each fritter, produce 64 coefficients: 1 dCcoefficient and 63 aCcoefficient, and right dCcoefficient and aCthe different quantization table of coefficient quantizes; To quantize after dCcoefficient carries out Differential pulse code modulation coding and entropy code, and after quantizing aCcoefficient then adopts RLC to encode and entropy code; Finally in conjunction with JPEG specify mark by certain form the bit stream produced after picture headers information and entropy code by byte-sized writing in files;
Note: RGB represents three kinds of color: R and represents red, and G represents green, and B represents blue; Y represents luminance component, U and V all represents color difference components, and dct transform represents discrete cosine transform, dCcoefficient represents DC coefficient, aCcoefficient represents ac coefficient, and RLC represents distance of swimming compressed encoding;
The JPEG coding stream sequence obtained mainly comprises two-part information: File header information and image compression data , the wherein supplementary needed for JPEG encoding and decoding such as the main include file size of File header information, quantization table, Huffman code tables, and image compression data is original image data by data after compressed encoding, it is expressed as:
(1)
Wherein, represent image ? (sample format of YUV is 4:1:1 to block, for Y-component block is then made up of 48 X 8 pieces, U and V component is made up of 18 X 8 pieces) through entropy code compression after data structure, ( ) represent that 16 X 16 pieces is at image the position of middle horizontal ordinate, represent image height, ( ) represent that 16 X 16 pieces is at image the position of middle ordinate, represent image wide, represent the bit stream after block Y-component entropy code, block size 8 X 8, , represent the position that block puts in order, represent the bit stream after U and V component entropy code respectively;
(1-2), basis data structure, extract image ? block Y's dCcoefficient bit stream and aCcoefficient bit stream; If image ? block Y-component dCbit stream is , it is expressed as:
(2)
Wherein, represent image the in block after block Y-component entropy code dCcoefficient bit stream, , represent the position that block puts in order;
If image ? block Y-component aCbit stream is , it is expressed as:
(3)
Wherein, ( ) represent image the in block after block Y-component entropy code aCcoefficient bit stream, represent the position that block puts in order;
From formula (2) (3), ( ) by with form, it is expressed as:
(4)
When time fixing, be actually the bit stream after 64 DCT coefficient entropy codings corresponding to 8 X 8 size block, wherein contain 1 dCcoefficient and 63 aCcoefficient, dCcoefficient and aCcoefficient coding structure is identical, all has two parts to form: Huffman code stream and VLI code stream;
Wherein, the Huffman encoding of Huffman representation, be a kind of harmless elongated compaction coding method, VLI is variable length integer coding;
When time, bitstream structure expression formula is:
(5)
Wherein, represent image ? block in component dCcoefficient Huffman code stream, represent image ? block in component dCcoefficient VLI code stream, utilizes image y-component in header file information dCcoefficient Huffman code table, the readable taking-up of the picture material owner bit stream, then right bit stream is decoded, and obtaining should dCcoefficient is group number in VLI code table, and its numerical values recited equals actual storage binary value bit stream length, reads out bit stream, obtains dCthe numerical value of coefficient before entropy code;
When time, bitstream structure expression formula is:
(6)
Wherein, represent in aCcoefficient individual coding unit corresponding Huffman code stream, represent the number of appearance zero continuously in Run-Length Coding, represent continuous zero rear first non-zero values corresponding group number in VLI code table, represent in aCcoefficient individual coding unit in corresponding VLI code stream, represent in aCthe number of coefficient Run-Length Coding unit, ; Utilize image y-component in header file information aCcoefficient Huffman code table, the readable taking-up of the picture material owner bit stream, then right bit stream is decoded, and obtaining should aCcoefficient in VLI code table group number (its value equals bit stream length, with reference to figure 6), read out bit stream;
By that analogy, read out further , can image be obtained ? block Y-component dCbit stream with aCbit stream ;
(1-3), according to (1-2) step, the picture material owner is from image readable taking-up the bit stream of U, V component corresponding to block , different unlike the Huffman code table of U, V component Huffman code table used and Y-component from the bit stream reading Y-component;
(1-4), based on (1-2) and (1-3) step, the picture material owner can obtain whole sub-picture middle Y, U, V component dCbit stream and aCbit stream; For image 's dCbit stream and aCbit stream, takes different encryption methods, can obtain ciphertext graph picture , specific as follows:
The picture material owner is by whole sub-picture middle Y, U, V component dCbit stream is combined into a new bit stream sequence , its expression formula is as follows:
(7)
Wherein, , , represent image respectively middle Y, U, V component dCbit stream sequence, ( ) represent bit stream the binary value of position;
If image corresponding bit stream length is , the method pair that the picture material owner adopts stream cipher XOR to encrypt be encrypted, namely according to encryption key producing a string length is pseudo-random binary bit stream with carry out xor operation, generating length is encryption after bit stream , xor operation meets following formula:
(8)
Namely , obtain ciphertext bit stream ;
For image middle Y, U, V component aCbit stream takes different encryption technologies and scrambling encryption method, extracts image according to formula (6) block bitstream structure in all 8 X 8 pieces of Y, U, V components aCcoefficient block bit stream, and respectively block bit stream Connecting groups is synthesized new bit stream sequence, be set to , , , expression formula is as follows respectively:
(9)
(10)
(11)
Wherein, represent image Y-component the block, block size 8 X the 8,1st aCcode stream section corresponding to coefficient coding unit, with reference to (1-2) step, represent the 1st aCbit stream part after coefficient coding unit Huffman encodes, represent expression the 1st aCbit stream part after coefficient coding unit VLI encodes, ( ) represent the position of 8 X 8 pieces horizontal ordinate in the y component, ( ) represent the position of 8 X 8 pieces ordinate in the y component;
represent that image U divides flow control block (block size 8 X 8) the 1st aCcode stream section corresponding to coefficient coding unit, with reference to (1-2) step, represent the 1st aCbit stream part after coefficient coding unit Huffman encodes, represent expression the 1st aCbit stream part after coefficient coding unit VLI encodes, ( ) represent the position of 8 X, 8 pieces of horizontal ordinates in U component, ( ) represent the position of 8 X, 8 pieces of ordinates in U component; Symbol implication in formula (11) is similar with formula (10);
Separately, image the size of corresponding Y, U, V component is respectively , ;
With ( ) code stream section integrally unit, the picture material owner uses encryption key right middle comprised code stream section carries out scrambling encryption, namely upsets code stream section exists position in bit stream, Y-component after note encryption aCbit stream is ; To U, V component aCbit stream , take identical encryption method to be encrypted, obtaining ciphertext bit stream is , , key used is respectively , ;
(1-5), according to the format standard of JPEG, by image header information and encryption after Y, U, V component bit stream carry out reconnecting combination, can image be obtained corresponding encrypted image bit stream sequence , its expression formula is:
(12)
Wherein, represent image header information, have neither part nor lot in encryption, its bit stream remains unchanged before encryption afterwards; represent length information, comprise bit stream , , , length and JPEG cataloged procedure in the figure place of filler, value exists with the form of bit stream;
Separately, mark in jpeg format and start with hexadecimal FF, after with the byte of a non-zero, when reading jpeg file in (1-1) step, if FF byte 0 below, then deletion value is this byte of 0, in this operation obtain encrypted image bit stream compares original image bit stream length reduce to a certain extent, ensure that the file size after encryption does not increase, even file size most image reduced on the contrary;
(1-6) other image encryption process, in image data base all and image cryptographic operation the same, thus obtain encryption after ciphertext image data base, finally ciphertext image data base is submitted to retrieval service supplier and carries out store and management.
3. according to claims 1 describe based on the histogrammic ciphertext domain image search method of jpeg image ac coefficient, it is characterized in that: the concrete operation step of described step (2) authorized user query image process is as follows:
(2-1), according to both sides' rendezvous protocol, need the user of inquiry service to obtain picture material supplier and authorize, become authorized user ;
(2-2), user the image of inquiry will be needed submit to image contents service person to be encrypted, the picture material owner adopts step (1-1), (1-2), (1-3), (1-4), (1-5) by user the plaintext image submitted to be encrypted to ciphertext graph picture , finally ciphertext graph picture with corresponding encryption key , , , return to user ;
(2-3), user by ciphertext graph picture submit to retrieval service supplier to retrieve, and the number that need return the most close image of announcement, wait for result for retrieval simultaneously;
(2-4), user the ciphertext graph picture that deciphering retrieval service supplier returns, if the ciphertext graph picture returned is , user read ciphertext graph picture , obtain image according to step (1-5) header information and compressed data information, reference formula (12); Image is obtained according to jpeg marker header information , and do not do extra process ;from image is obtained in bit stream in , , , the length of bit stream, is partitioned into according to respective bit stream length , , , four section streams
Reference formula (8), utilizes well-known key produce pseudo-random binary bit stream with carry out xor operation to decrypt , for , utilize key scramble the position of middle code stream section, recovers code stream section original position, namely decrypt bit stream; By that analogy, key is utilized , decrypt , bit stream;
Decompose according to step (1-4) , , , , obtain image middle Y, U, V component 8 X 8 pieces is corresponding dCwith aCbit stream, utilizes formula (1) to be combined into image the packed data of entropy code , combining image header information namely image is obtained bit stream , its expression formula is:
(13)
Separately, according to JPEG coding standard, image bit stream is by byte writing in files, if occur, byte value is FF, and is not JPEG regulation mark, then need fill a value after FF byte is the byte of 0, namely fill in step (1-5) deleted fall value be the byte of 0, thus decrypt ciphertext graph picture original image if return how secondary ciphertext graph picture, be all decrypted operation recovery according to step (2-4) and go out corresponding original plaintext image.
4. according to claims 1 describe based on the histogrammic ciphertext domain image search method of jpeg image ac coefficient, it is characterized in that: the concrete operation step that described step (3) retrieval service supplier retrieves ciphertext image process is as follows:
(3-1), retrieval service supplier read authorized user the ciphertext graph picture to be checked submitted to , the plaintext image of its correspondence is , obtain in ciphertext graph picture in , , three section streams, acquisition process refer step (2-4), in conjunction with Huffman code table and VLI code table, extracts ciphertext graph picture y, U, V tri-components all aCcoefficient, and utilize mathematical statistics means to obtain image y, U, V tri-components aCthe histogram that coefficient is corresponding , , ;
Separately, adopt to use the same method and ask all images in ciphertext database corresponding aCcoefficient histogram;
(3-2), ciphertext graph picture is calculated one by one and ciphertext graph is as the Histogram distance between all images of lane database, this distance is the measure of two width image similarity, and between distance less image, similarity is higher, otherwise then similarity is lower; The present invention adopts ratio of precision euclidean distance is higher relative distance, i.e. city block distance, its expression formula is as follows:
(14)
Wherein, for image to be checked, for image arbitrary in image data base, for query image middle Y-component aCcoefficient is frequency, for query image middle Y-component aCcoefficient is frequency; One secondary jpeg image is made up of Y, U, V tri-components, utilizes formula (14) can try to achieve U, V two components respectively aCthe distance of coefficient histogram , , cumulative three component distances are as dimensioned plan picture and image the final distance of similarity , expression formula is as follows:
(15)
Wherein, respectively , , weight, take exhaustive method to make , , weight optimum in formula (15);
Formula (14) (15) are utilized to obtain ciphertext graph picture one by one and ciphertext graph is as Histogram distance (comprising Y, U, V component) between lane database all images;
(3-3), to the ciphertext graph picture required by (3-2) step and ciphertext graph sorts from small to large as the Histogram distance between all images of lane database, forward preferentially the returning of sequence number, specifically returns ciphertext graph picture in how many auxiliary image data storehouse by authorized user tell retrieval service supplier in advance.
CN201410591470.XA 2014-10-29 2014-10-29 Ciphertext domain image search method based on jpeg image ac coefficient histogram Expired - Fee Related CN104331448B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201410591470.XA CN104331448B (en) 2014-10-29 2014-10-29 Ciphertext domain image search method based on jpeg image ac coefficient histogram

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201410591470.XA CN104331448B (en) 2014-10-29 2014-10-29 Ciphertext domain image search method based on jpeg image ac coefficient histogram

Publications (2)

Publication Number Publication Date
CN104331448A true CN104331448A (en) 2015-02-04
CN104331448B CN104331448B (en) 2018-04-06

Family

ID=52406175

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201410591470.XA Expired - Fee Related CN104331448B (en) 2014-10-29 2014-10-29 Ciphertext domain image search method based on jpeg image ac coefficient histogram

Country Status (1)

Country Link
CN (1) CN104331448B (en)

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106791882A (en) * 2016-12-05 2017-05-31 Tcl集团股份有限公司 A kind of coding encrypting method and device of image
CN107979711A (en) * 2017-11-28 2018-05-01 上海大学 Based on the method for optimizing distortion modification prevention hided transmission
CN108182220A (en) * 2017-12-25 2018-06-19 重庆邮电大学 Image search method based on privacy of user protection in Cloud Server
CN109151486A (en) * 2018-09-06 2019-01-04 西南交通大学 The jpeg image bit stream encrypted domain reversible data concealing method of large capacity
CN111325807A (en) * 2020-02-24 2020-06-23 南京信息工程大学 Encryption and feature extraction method based on JPEG image
CN112231508A (en) * 2020-10-15 2021-01-15 中山大学 Encrypted JPEG image retrieval method based on content
CN112417190A (en) * 2020-11-27 2021-02-26 暨南大学 Retrieval method and application of ciphertext JPEG image
CN112434180A (en) * 2020-12-15 2021-03-02 暨南大学 Ciphertext JPEG image retrieval method based on DCT coefficient value replacement and attention network
CN112528064A (en) * 2020-12-10 2021-03-19 西安电子科技大学 Privacy-protecting encrypted image retrieval method and system
CN115378590A (en) * 2022-10-27 2022-11-22 国网浙江义乌市供电有限公司 Energy data safe storage method and system based on block chain
CN115733617A (en) * 2022-10-31 2023-03-03 支付宝(杭州)信息技术有限公司 Biological characteristic authentication method and system
CN117676043A (en) * 2024-02-01 2024-03-08 无锡沐创集成电路设计有限公司 Pseudo-color image transmission method, system, equipment and storage medium

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000106680A (en) * 1998-09-28 2000-04-11 Nec Eng Ltd Image compressing and expanding device
JP2002094924A (en) * 2000-09-12 2002-03-29 Fuji Photo Film Co Ltd Image data recorder and method, and image data reproducing device and method
CN103744976A (en) * 2014-01-13 2014-04-23 北京工业大学 Secure image retrieval method based on homomorphic encryption
CN104008174A (en) * 2014-06-04 2014-08-27 北京工业大学 Privacy-protection index generation method for mass image retrieval

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000106680A (en) * 1998-09-28 2000-04-11 Nec Eng Ltd Image compressing and expanding device
JP2002094924A (en) * 2000-09-12 2002-03-29 Fuji Photo Film Co Ltd Image data recorder and method, and image data reproducing device and method
CN103744976A (en) * 2014-01-13 2014-04-23 北京工业大学 Secure image retrieval method based on homomorphic encryption
CN104008174A (en) * 2014-06-04 2014-08-27 北京工业大学 Privacy-protection index generation method for mass image retrieval

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
文昌辞等: "JPEG彩色图像加密新算法", 《系统工程与电子技术》 *
朱旭东等: "云计算环境下加密图像检索", 《西安电子科技大学(自然科学版)》 *

Cited By (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106791882B (en) * 2016-12-05 2019-12-13 Tcl集团股份有限公司 image coding and encrypting method and device
CN106791882A (en) * 2016-12-05 2017-05-31 Tcl集团股份有限公司 A kind of coding encrypting method and device of image
CN107979711A (en) * 2017-11-28 2018-05-01 上海大学 Based on the method for optimizing distortion modification prevention hided transmission
CN107979711B (en) * 2017-11-28 2019-07-23 上海大学 Based on the method for optimizing distortion modification prevention hided transmission
CN108182220A (en) * 2017-12-25 2018-06-19 重庆邮电大学 Image search method based on privacy of user protection in Cloud Server
CN109151486B (en) * 2018-09-06 2020-10-09 西南交通大学 JPEG image bit stream encryption domain reversible data hiding method
CN109151486A (en) * 2018-09-06 2019-01-04 西南交通大学 The jpeg image bit stream encrypted domain reversible data concealing method of large capacity
CN111325807A (en) * 2020-02-24 2020-06-23 南京信息工程大学 Encryption and feature extraction method based on JPEG image
CN111325807B (en) * 2020-02-24 2023-11-24 南京信息工程大学 Encryption and feature extraction method based on JPEG image
CN112231508A (en) * 2020-10-15 2021-01-15 中山大学 Encrypted JPEG image retrieval method based on content
CN112231508B (en) * 2020-10-15 2022-06-24 中山大学 Encrypted JPEG image retrieval method based on content
CN112417190A (en) * 2020-11-27 2021-02-26 暨南大学 Retrieval method and application of ciphertext JPEG image
CN112528064B (en) * 2020-12-10 2022-12-13 西安电子科技大学 Privacy-protecting encrypted image retrieval method and system
CN112528064A (en) * 2020-12-10 2021-03-19 西安电子科技大学 Privacy-protecting encrypted image retrieval method and system
CN112434180A (en) * 2020-12-15 2021-03-02 暨南大学 Ciphertext JPEG image retrieval method based on DCT coefficient value replacement and attention network
CN112434180B (en) * 2020-12-15 2024-03-15 暨南大学 Ciphertext JPEG image retrieval method based on DCT coefficient value replacement and attention network
CN115378590A (en) * 2022-10-27 2022-11-22 国网浙江义乌市供电有限公司 Energy data safe storage method and system based on block chain
CN115733617A (en) * 2022-10-31 2023-03-03 支付宝(杭州)信息技术有限公司 Biological characteristic authentication method and system
CN115733617B (en) * 2022-10-31 2024-01-23 支付宝(杭州)信息技术有限公司 Biological feature authentication method and system
CN117676043A (en) * 2024-02-01 2024-03-08 无锡沐创集成电路设计有限公司 Pseudo-color image transmission method, system, equipment and storage medium

Also Published As

Publication number Publication date
CN104331448B (en) 2018-04-06

Similar Documents

Publication Publication Date Title
CN104331448A (en) Encrypted domain image retrieval method based on JPEG (Joint Picture Expert Group) image alternating current coefficient histograms
Qian et al. New framework of reversible data hiding in encrypted JPEG bitstreams
Cheng et al. Markov process-based retrieval for encrypted JPEG images
Cheng et al. Encrypted JPEG image retrieval using block-wise feature comparison
US20160294781A1 (en) Partial or complete image obfuscation and recovery for privacy protection
CN104144343A (en) Digital image compressing, encrypting and encoding combined method
Liang et al. Huffman-code based retrieval for encrypted JPEG images
Rashid et al. Secure image deduplication through image compression
Cheng et al. AC-coefficient histogram-based retrieval for encrypted JPEG images
JP2017535123A (en) Encoder, decoder and method using partial data encryption
WO2005078989A1 (en) Repetition coded compression for encrypting highly correlated data
CN104038767A (en) Encryption domain H.264/AVC (Advanced Video Coding) video data hiding method compatible with CABAC (Context-based Adaptive Binary Arithmetic Coding)
CN111325807B (en) Encryption and feature extraction method based on JPEG image
CN108322750A (en) Encrypted domain HEVC video data hidden methods based on residual error coefficient modulation
Kalita et al. A comparative study of steganography algorithms of spatial and transform domain
Iida et al. An image identification scheme of encrypted jpeg images for privacy-preserving photo sharing services
Tang et al. OPPR: An outsourcing privacy-preserving JPEG image retrieval scheme with local histograms in cloud environment
CN108600573B (en) Ciphertext JPEG image retrieval method based on tree-shaped BoW model
Ye et al. Secure multimedia big data sharing in social networks using fingerprinting and encryption in the jpeg2000 compressed domain
Cheng et al. Toward privacy-preserving JPEG image retrieval
Li et al. A privacy-preserving scheme for JPEG image retrieval based on deep learning
Rodrigues et al. A selective encryption for heterogenous color JPEG images based on VLC and AES stream cipher
Reddy et al. Steganpeg steganography+ JPEG
CN101431674A (en) Video monitoring system
Devi et al. Secured covert color image transmission using secret fragment visible mosaic image and reversible color transformation technique

Legal Events

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

Granted publication date: 20180406

Termination date: 20201029