Arjun Nichal, Dr. Shraddha Deshpande / International Journal of Engineering Research and Applications                     ...
Arjun Nichal, Dr. Shraddha Deshpande / International Journal of Engineering Research and Applications                     ...
Arjun Nichal, Dr. Shraddha Deshpande / International Journal of Engineering Research and Applications                     ...
Arjun Nichal, Dr. Shraddha Deshpande / International Journal of Engineering Research and Applications                     ...
Arjun Nichal, Dr. Shraddha Deshpande / International Journal of Engineering Research and Applications                     ...
Upcoming SlideShare
Loading in …5



Published on

IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit

  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide


  1. 1. Arjun Nichal, Dr. Shraddha Deshpande / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue 4, June-July 2012, pp.751-755 A High Capacity Data Hiding Method for JPEG2000 Compression System Arjun Nichal*, Dr. Shraddha Deshpande** *(Department of Electronics, Walchand College of Engineering, Sangli, Maharashtra 416415, India ** (Department of Electronics, Walchand College of Engineering, Sangli, Maharashtra 416415, IndiaABSTRACT Data hiding is an important branch of method is used [5]-[6]. Redundancy evaluation methodinformation security. Imperceptibility and Hiding determines embedding depth adaptively for increasingCapacity are very important aspects for efficient secret hiding capacity. A candidate embedding point will becommunication. It is necessary to increase hiding removed if evaluated quantization redundancy is less thancapacity for JPEG2000 baseline system because two. For security purpose secret message is encrypted byavailable redundancy is very limited. In this paper using encryption key. At decoder side exact inverseRedundancy Evaluation method is used for increasing procedure is used for extraction of secret message.hiding capacity. This method determines embeddingdepth adaptively for increasing hiding capacity. Large I. SECRET MESSAGE EMBEDDING ALGORITHMquantity of data is embedded into bitplanes, but at the This message embedding algorithm uses redundancycost of slightly change in Peak Signal to Noise Ratio evaluation method for increasing hiding capacity.(PSNR). This method easily implemented in JPEG2000 Quantized secret message embedded wavelet coefficientscompression encoder and produced stego stream compressed by using JPEG2000 compression baselinedecoded normally at decoder. Simulation result shows system. Secret message is encrypted by using secretthat this method is secure and increases hiding capacity. encryption key. Same secret encryption key is used at decoder side for extraction of secret messageKeywords - Secret communication, JPEG2000, Data DWT Quantization Redundancyhiding, Information security, PSNR. Evaluation Cover ImageINTRODUCTION Key Processing Information hiding in digital images has drawn Secret Image Blockmuch attention in recent years. Secret message encrypted Messageand embedded in digital cover media. The redundancy of Secret Key Embedding Blockdigital media as well as characteristics of human visualsystem makes it possible to hide secret messages. Twocompeting aspects are considered while designing EBCOTinformation hiding scheme 1) Hiding capacity and 2) EncoderImperceptibility. Hiding capacity means maximum payload.Imperceptibility means keeping undetectable [1]. Codestrem A least significant bits (LSB) substitution methodis widely used for hiding data in digital images. This Fig. 1 Secret message embedding block diagrammethod widely used because of large capacity and easyimplementation. This kind of secret data embedding 1.1 Discrete Wavelet Transformapproach carried out in image pixel and quantized discrete Cover image is decomposed by using discrete waveletcosine transform (DCT) [2]-[3]. In JPEG compression transform up to certain level.system secure data hiding scheme achieved by modifyingquantized DCT coefficients. A DCT domain data hiding 1.2 Quantizationscheme can be applied in JPEG very conveniently. A Uniform scalar quantizer is used for quantization purpose.JPEG2000 international coding standard is based on 𝑦 𝑏 [𝑛]discrete wavelet transform. Some data hiding schemes 𝑞 𝑏 𝑛 = 𝑠𝑖𝑔𝑛(𝑦 𝑏 [𝑛]) (1) ∆𝑏cannot fitted to JPEG2000 compression system directly. Allsecret data will be destroyed because of truncating Here 𝑦 𝑏 [𝑛] denotes the sample of subbands, while 𝑞 𝑏 [𝑛]operation if it is embedded into lowest bitplanes [4]. For denotes the quantization indices and ∆ 𝑏 denotes the stepJPEG2000 compression standard limited redundancy and size.bitstream truncation makes it difficult to hide information.To overcome these two problems redundancy evaluation 751 | P a g e
  2. 2. Arjun Nichal, Dr. Shraddha Deshpande / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue 4, June-July 2012, pp.751-7551.3 Redundancy Evaluation 1.4 Key Processing Block The redundancy of uniform quantization is In this step secret image is encrypted by usingdetermined according to the visual masking effect and secret key. Encryption process hides the contents of thebrightness sensitivity of human visual system. The information whereas stegnography hides existence of thecalculation of self- contrast masking effect, neighborhood information. Suppose 𝑚 𝑖 is a secret data and 𝑛 𝑖 is a secretmasking effect and local brightness weighting factor is key, then final encrypted data will be,calculated as follows. In the first step self-contrast maskingeffect is calculated, 𝐸𝑛𝑐𝑟𝑦𝑝𝑡𝑒𝑑 𝑑𝑎𝑡𝑎 = 𝑚 𝑖 ⊕ 𝑛 𝑖 (8) ⍺ 1.5 Message Embedding Block 𝑦 𝑖 = 𝑠𝑖𝑔𝑛( 𝑥 𝑖 ) 𝑥 𝑖 . 𝛥 𝑖 (2) In this message embedding block message isWhere 𝑥 𝑖 is the quantized wavelet coefficient with bits embedded adaptively in quantized wavelet coefficient. Firstlower than highest no-zero bit are replaced by zeros. 𝛥 𝑖 is threshold assigned for embedding process. Waveletquantization step. Parameter ⍺ takes value between 0 and 1. coefficients greater than given threshold are chosen asResult of first step is𝑦 𝑖 . In second step neighborhood candidate embedding point. Threshold calculated asmasking is calculated follows. 𝑦𝑖 𝑧𝑖 = (3) 𝑇𝑕𝑟𝑒𝑠𝑕𝑜𝑙𝑑 = 2 𝑛 𝑤𝑕𝑒𝑟𝑒 𝑛 = 4,5,6,7 (9) 1+(𝑎 𝑘⋳ 𝑛𝑒𝑖𝑔 𝑕 𝑏𝑜𝑟 𝑕 𝑜𝑜𝑑 𝑥 𝑘 𝛽 )/ ɸ 𝑖 Next step is to calculate lowest embed allowed bitplane. The symbol 𝑥 𝑘 denotes the neighboring wavelet This can be calculated by using following formula.coefficients greater than or equal to 16 and its bits lowerthan highest no-zero are replaced by zeros. Symbol ɸ 𝑖 is log (𝑡𝑕𝑟𝑒𝑠 𝑕𝑜𝑙𝑑 )number of wavelet coefficient in neighborhood. Parameter 𝐿𝑜𝑤𝑒𝑠𝑡 𝑒𝑚𝑏𝑒𝑑 𝑏𝑖𝑡𝑝𝑙𝑎𝑛𝑒 = −1 (10) log (2)β assumes value between 0 and 1. Parameter 𝑎 is anormalization factor and having constant value 𝑎 = Bitplanes lower than lowest embed allowed(10000/2 𝑑−1 ) 𝛽 . d is a bit depth of image component. In bitplanes are truncated during compression process.third step weighting factor about brightness sensitivity is Now suppose value of threshold is 16. Four waveletcalculated. coefficients A, B, C and D having values 33, 65, 9 and 128 respectively. Coefficient C cannot chosen for embedding 2 − 𝐿 𝑙, 𝑖, 𝑗 , 𝑖𝑓 𝐿 𝑙, 𝑖, 𝑗 < 1 process because its value is less than threshold. Lowest Ʌ 𝑙, 𝑖, 𝑗 = (4) 𝐿 𝑙, 𝑖, 𝑗 , 𝑂𝑡𝑕𝑒𝑟𝑤𝑖𝑠𝑒 embed allowed bitplane is 3. Now after calculation of redundancy, result for wavelet coefficients A, B and D as 1 𝑖 𝑗 follows. 𝐿 𝑙, 𝐼, 𝐽 = 1 + 𝐼𝑘 1+ ,1 + (5) 128 0 2 𝑘−1 2 𝑘−1Ʌ 𝑙, 𝑖, 𝑗 is a local brightness weighting factor. The symbol 𝑟 𝐴 = 1.59 𝑟 𝐵 = 2.60 𝑟 𝐷 = 12.5𝐼 𝑙 𝜃 denotes the sunband at resolution level 𝑙 ⋳ 0,1, … . . 𝑘and with orientation 𝜃 ⋳ 𝐿𝐿, 𝐿𝐻, 𝐻𝐿, 𝐻𝐻 . The symbol𝐼 𝑙 𝜃 (𝑖, 𝑗) denotes the wavelet coefficient located at (𝑖, 𝑗) insubband 𝐼 𝑙 𝜃 . The level of discrete wavelet decomposition is 𝑘. The result of next step is 𝑧𝑖 𝑧 ′𝑖 = (6) Ʌ(𝑙,𝑖,𝑗 )Final quantization redundancy is calculated by usingfollowing equation. 𝑥 𝑟𝑖 = ′𝑖 (7) 𝑧𝑖 The redundancy of wavelet coefficient 𝑥 𝑖 can bemeasured by 𝑟𝑖 , for avoiding degradation of image we usewavelet coefficients with 𝑟𝑖 not less than 2 to carry Fig. 2 Candidate Embedding Pointsmessage bits.The rules of adjusting embedding points andintensity is as follows. Value of 𝑟 𝐴 is less than 2 . Wavelet coefficient 𝐴 cannot be used for embedding process. 1) If 𝑟𝑖 < 2, then embedding point should be removed. Value of 𝑟 𝐵 is greater than 2. 21 < 2.60 < 22 so the 2) If 2 𝑛 ≤ 𝑟𝑖 < 2 𝑛+1 then embedding capacity of this embedding capacity of coefficient 𝐵 is 1 bit. point is determined by 𝑛 bits.The final 𝑛 𝑖 is the adaptive embedding depth for each Value of 𝑟 𝐷 is greater than 2. 23 < 12.5 < 24 so thequantized wavelet coefficient 𝑥 𝑖 . embedding capacity of coefficient 𝐷 is 3 bits. In next step encrypted bits are embedded into selected wavelet coefficients adaptively. 752 | P a g e
  3. 3. Arjun Nichal, Dr. Shraddha Deshpande / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue 4, June-July 2012, pp.751-755 III. QUALITY PARAMETERS For comparing With Redundancy Evaluation Method and Without Redundancy Evaluation Method we have considered various quality parameters such as Compression Ratio (CR), Peak Signal to Noise Ratio (PSNR) and Embedding Capacity (EC). 3.1 Compression Ratio (CR) The compression ratio is calculated as the ratio of number of bits required to represent original image to the number of bits required to represent compressed codestream. Number of bits required to represent original image Fig. 3 Adjusted embedding points and intensity 𝐶𝑅 = Number of bits required to (12)1.6 Embedded Block Coding and Optimized Truncation represent compressed codestream(EBCOT) Encoder After message embedding process message 3.2 Peak-Signal to Noise Ratio (PSNR)embedded quantized subbands are encoded by using The PSNR is calculated by using following formula.Embedded Block coding and optimized truncation 𝑀𝐴𝑋 2(EBCOT) encoder. EBCOT encoder is used in JPEG2000 𝑃𝑆𝑁𝑅 = 10𝑙𝑜𝑔10 𝐼 (13) 𝑀𝑆𝐸baseline system for encoding purpose. Three coding passes 𝑚 −1 𝑛−1are used in EBCOT encoder 1) Significance propagation 1 2 𝑀𝑆𝐸 = 𝐼 𝑖, 𝑗 − 𝐾(𝑖, 𝑗)pass 2) Magnitude refinement pass 3) Clean-up pass. These 𝑚𝑛 𝑖=0 𝑗 =0three coding passes are used for encoding purpose. Finaloutput of this EBCOT encoder block is one dimensional Where 𝑀𝐴𝑋 𝐼 is maximum possible pixel value ofcodestream. image 𝐼 𝑖, 𝑗 . 𝑀𝑆𝐸 is a mean square error. 𝐼 𝑖, 𝑗 is a original cover image. 𝐾(𝑖, 𝑗) is a reconstructed coverII. SECRET MESSAGE EXTRACTION ALGORITHM image. 𝑚 is number of rows and 𝑛 is number of columns.Extraction process is simply inverse process to that ofembedding process. 3.3 Embedding Capacity (EC) Embedding capacity is calculated by using followingCodestream formula. 𝑛 𝑚 Codestream Redundancy decoder Evaluation 𝐸𝐶 = 𝑛 𝑖, 𝑗 𝑖=1 𝑗 =1 Message Extraction 𝑖𝑓 x 𝑖, 𝑗 > 𝑇𝑕𝑟𝑒𝑠𝑕𝑜𝑙𝑑 (14) x 𝑖, 𝑗 is a quantized wavelet coefficient matrix. Secret Key Extracted IDWT IV. EXPERIMENTAL RESULTS AND DISCUSSIONS Data In this paper binary secret image is embedded into grayscale cover image. Image into image steganography scheme is to be carried out. Some standard grayscale Rec. Secret Rec. cover images are used as a cover images. With Redundancy Image Image Evaluation Method and Without Redundancy Evaluation Method are compared on the basis of various performance Fig. 4 Secret message extraction algorithm parameters such as Compression Ratio (CR), Peak Signal toCodestream is the input for secret message extraction Noise Ratio (PSNR) and Embedding Capacity (EC).algorithm. Output of the codestream decoder is a subbands. Lena image having size 512*512 is used for processing.After codestream decoder redundancy is to be evaluated asdescribed in secret message embedding algorithm. By usingquantization redundancy matrix 𝑟𝑖 embedded secret bits areextracted from subbands. Extracted secret bits are encryptedbits. So original secret image recovered by using followingformula. 𝑚 𝑖 = 𝐸𝑛𝑐𝑟𝑦𝑝𝑡𝑒𝑑 𝑑𝑎𝑡𝑎 ⊕ 𝑛 𝑖 (11)Where 𝑚 𝑖 is a recovered secret image and 𝑛 𝑖 is encryptionkey. Encryption key used at embedding operation andextraction operation are same. Inverse discrete wavelet (a) (b)transform (IDWT) is used for recovering cover image Fig. 5 (a) Original Lena Image, (b) Secret Image 753 | P a g e
  4. 4. Arjun Nichal, Dr. Shraddha Deshpande / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue 4, June-July 2012, pp.751-755 Now with compression rate 0.5, Decomposition 14000level 5 and threshold 32 determined embedding capacity of With Red. EvaluationLena image is 5575 bits. 5476 bits embedded in Cover 12000image Lena with compression ratio 15.9959. At decoder Without Red. Evaluationside same parameters used for extraction of secret message. 10000After decoding process value of PSNR is 34.2839db. Onlythe first, Second and third decomposition levels are used to 8000embed secret messages. The forth and fifth decompositionlevel contains important low frequency components that 6000cannot modified to hide secret messages. 4000 2000 0 0 0.2 0.4 0.6 0.8 1 . Fig. 7 Compression rate Vs Embedding Capacity Compression rate Vs embedding capacity is plotted for both With and Without Redundancy Evaluation method. (a) (b) Fig. 6 (a) Reconstructed Lena Image, (b) Retrieved 16000 Secret Image With Red. Evaluation 14000 Table 1: Results for Lena image with diff. compression Without Red. Evaluation rate and common threshold = 32 12000 Without With Redundancy 10000 C. Redundancy C. EvaluationRate Evaluation Ratio 8000 EC NBE PSNR EC NBE PSNR 60000.2 1482 1444 30.973 741 676 31.217 39.974 40000.4 4092 3844 33.522 2045 1936 34.322 19.993 20000.6 7129 7056 35.065 3564 3364 36.154 13.330 00.8 10094 10000 36.328 5049 4900 37.446 9.998 16 32 64 1281 13021 12996 37.263 6515 6400 38.414 7.998 . Fig. 7 Threshold Vs Embedding Capacity Threshold Vs Embedding capacity (EC) plotted for bothC. Rate is a compression rate, EC is a embedding Capacity, With as well as Without Redundancy Evaluation Method.NBE is number of bits embedded, PSNR is Peak Signal toNoise Ratio and C. Ratio is a Compression Ratio. V. CONCLUSION Table 2: Results of Lena Image with diff. Thresholds Results produced with With Redundancy and common compression rate = 0.5. Evaluation method yields in large embedding capacity than Without Redundancy Evaluation method. Without changing Threshold EC of With RE EC of Without RE much image quality maximum secret data is to embedded. Redundancy evaluation method increases 16 13452 6725 Embedding capacity (EC) at the cost of slight change in Peak Signal to Noise Ratio (PSNR). 32 5574 2787 This effort gives comprehensive study of image 64 1915 957 steganography for JPEG2000 baseline system with variety of quality parameters. 128 420 210EC of With RE means Embedding capacity With REFERENCESredundancy evaluation and EC of Without RE means [1] N. Proves and P. Honeyman, “Hide and seek: AnEmbedding Capacity of Without Redundancy Evaluation. introduction to Steganography,” IEEE Security and Privacy Mag., vol. 1, no. 3, pp. 32-44, 2003. 754 | P a g e
  5. 5. Arjun Nichal, Dr. Shraddha Deshpande / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue 4, June-July 2012, pp.751-755[2] H. C. Wu, N. I. Wu, C. S. Tsai, and M. S. Hwang, “Image steganographic scheme based on pixel- value differencing and LSB replacement methods,” IEE Proc. Vis., Image , Signal Process., vol. 152, no.5, pp. 611–615, Oct. 2005.[3] C. K. Chan and L. M. Cheng, “Hiding data in images by simple LSB substitution,” Pattern Recognit., vol. 37, no. 3, pp. 469–474, 2004.[4] JPEG2000 Part 1: Final Committee Draft Version 1.0, ISO/IEC. FCD 15444-1, 2000.[5] P. C. Su and C. C. J. Kuo, “Steganography in JPEG2000 compressed images,” IEEE Trans. Consum. Electron., vol. 49, no. 4, pp. 824–832, Apr. 2003.[6] Liang Zhang, Haili Wang and Renbiao Wu, “A high capacity stegnography scheme for JPEG2000 baseline system,” IEEE Transactions on Image Processing, vol. 18, no. 8, August 2009. 755 | P a g e