A Novel Image Coding Algorithm Using DTT and SPIHT

DOI : 10.17577/IJERTV2IS120909

Download Full-Text PDF Cite this Publication

Text Only Version

A Novel Image Coding Algorithm Using DTT and SPIHT

A Novel Image Coding Algorithm Using DTT and SPIHT

Dr. P. Ramana Reddy, S. Siddeswara Reddy,

Associate professor in ECE, M. Tech,

JNTUACE, Ananthapuram, JNTUACE, Ananthapuram,


In recent years, most of the research activities are focused on wavelet based image coders rather than DCT based image coders. This is mainly attributed due to innovative strategies of data organization and representation of wavelet transformed coefficients. This paper presents a Discrete Tchebichef Transform (DTT) based hybrid image coder for image compression applications. In this coder DTT is coupled with set partitioning in hierarchical coding techniques (SPIHT).The compression and image reconstruction performance is compared with existing coders..

  1. Introduction

    Data compression has become an area of great interest in comparing due to the limited data storage and Network capabilities. In recent years, most of the research activities are focused on wavelet based image coders rather than DCT based image coders. This is mainly attributed due to innovative strategies of data organization and representation of wavelet transformed coefficients. The Discrete Cosine Transform (DCT) is a transform method on discrete signals and used in many signal processing applications including JPEG image compression standard.

    A new class of transform known as Discrete Tchebichef Transform (DTT) is another transform method introduced recently and based on the orthogonal version of Tchebichf polynomials. It has been shown that the DTT has a good energy compaction property, performs as well as the DCT for many images and performs better for certain classes of images. It is that the good energy Compaction property will allow describing major feat of images by relatively few numbers of values. The image compression based on the DTT will have a good image compression ratio and compress images as quickly as possible.DTT has been found excellent rate-distortion trade off like DCT

    and out performs DCT for image having high intensity graduations. Therefore we implemented to use DTT as a substitute for DCT in an image coder. The implemented image coder consists of DTT and SPIHT coding technique. The performance of this kind of coder is evaluated and is compared with DCT based image coders, JPEG, improved JPEG; it has found that the implemented coder best performs most of the coders.

  2. LL1










    Fig.1Illustigration of 3level wavelet Decomposition

    The image coder is similar to the wavelet transformed image across different scales by using a hierarchical tree structure.1-D set of samples are converted into the low-pass sub-band (Li) and high pass sub-band (Hi).The low pass sub-band represents a down sampled low resolution version of the original image. The high

    pass sub-band represents residual information of the

    original image, needed for the perfect reconstruction of the original image from the low pass sub-band.

    The 1-D wavelet transform can be extended to a two dimensional (2-D) wavelet transform using separable wavelet filters. With separable filters the 2-D transform can be computed by applying a 1-D transform to all the rows of the input, and then repeating on all of the

    Given Transformed coefficients T of the Two dimensional DTT of an N*N input f. The inverse Two dimensional DTT is defined as follows:

    f (x,y) = y)


    The example is repeated for a Three-Level (k=3)

    wavelet expansion in fig:1. In all of the discussion k represents the highest level of the wavelet transform.

    The 2-D sub-band decomposition is just an extension of 1-D sub-band decomposition. The entire process is carried out by executing 1-D sub-band decomposition twice, first in one direction (horizontal), then in the orthogonal (vertical) direction. For example, the low passes sub-band (Li) resulting from the horizontal direction is further decomposed in the vertical direction, leading to LLi and LHi sub-bands.

  3. Discrete Tchebichef Transform

    The Discrete Tchebichef Transform (DTT) is a linear transform with the kernel defined by the orthogonal version of Tchebichef polynomials. That the DTT has higher energy compactness that the DCT in images that have high illumination value variations such as artificial diagrams. The family of the DTT is also used in the field of image feature extraction.

    The DTT is defined as follows Given a two dimensional input f(x,y) of site N*N,the transformed coefficients, Tpq of the two dimensional discrete Tchebichef Transform is defined as follows

    f(x,y) The kernel is

  4. Image Compression Using DTT-SPIHT

    The implemented DTT-SPIHT embedded coder is shown in Fig.2.The input image is divided into non- overlapping 8*8 block. Each block is transformed using discrete tchebichef transform. The coefficients are arranged into 3 level wavelet pyramid structure. The coefficients are quantized by SPIHT coding algorithm. The next stage is to use entropy coding which is optional. The proposed algorithm does not use entropy coding. used by our algorithm. By incorporating arithmetic coding stages in our coder, additional 5-10 percentage compression can be achieved.

    Discrete Tchebichef Transform

    Discrete Tchebichef Transform

    Wavelet like coefficient

    Wavelet like coefficient

    8*8 block Extraction

    8*8 block Extraction

    input image

    Compressed image



    Perceptual weights

    Segmen tation



    Perceptual weights

    Segmen tation

    Fig:2. Block diagram of DTT-SPHIT Image coder

    The arrangement of 8*8 DTT coefficients in a 3 Level wavelet pyramid structure. After labeling 64 coefficients in each block, the parent child relationship



    is defined as follows: The parent of coefficient i is [i/4] for 1 i 63, while the set of four children associated with coefficient j is { 4j,4j+1,4j+2,4+3 } for 1 j 15.The DC coefficient 0 is the root of DTT coefficients

    tree, which has only three children: coefficients 1,2 and



    3.In the proposed structure, off springs corresponds to

    direct descendants in the same spatial location in the next finer band of the pyramid. A tree corresponds to a node having four children which always form a group

    of 2*2 adjacent pixels. In Fig.3 indicate that the index

    coefficients of other 8*8 blocks are grouped together so

    that the entire Image can form an overall 3-level pyramid structure. In the implemented decomposition

    method, we further decompose LL3 band into a 3- level pyramid so that, the coarsest level will be a 8×8 band

    .the overall level of decomposition is now fix. Then SPIHT encoding algorithm is applied to the overall structure.

    34 35 38 39

    0 1

    4 5

    16 17 20 21

    2 3

    6 7

    18 19 22 23

    8 9

    12 13

    24 25 28 29

    10 11

    14 15

    26 27 30 31

    32 33 36 37

    48 49 52 53

    50 51 54 55

    40 41 44 45

    56 57 60 61

    42 43 46 47

    58 59 62 63

    Fig.3. Rearrangement algorithm of 8×8 transformed coefficients.

    SPIHT algorithm keeps track of the state of sets by means of three lists, i.e., list of insignificant sets(LIS). List of significant pixels (LSP) and lists of insignificant pixels(LIP).The algorithm uses the following sets to code a bitmap effectively(i,j) is the set of coordinates of all offsprings of node (i,j), D(i,j) is set of coordinates of all descendents of node (i,j), L(i,j) is the set of coordinates defined as D(i,j)-O(I,j), H(I,j) is set of all three roots. The significance test of the a wavelet coefficient is defined as follows

    while marking as type B. otherwise, remove the node (I,j) from LIS. If the node belongs to type B, then output Sn(L(I,j))=1, append the fore direct subsequent nodes to the LIS as type A.

    • Refinement pass: For each element (i,j) in the LSP, output the n-th most significant bit except those added above.

    • Update: Decrement n by 1, and go to sorting pass.

  5. Simulation Results

    To evaluate the performance of hybrid image coding algorithm, experiments are conducted on Lena image. The size of image is 512×512.Input image to DTT- SPIHT algorithm is shown in fig .4.(a). After performing the DTT-SPIHT algorithm we obtained the compressed Lena image is shown in fig.4.(b). With compression ratio is 29.3166. The compressed Lena image is applied to inverse DTT-SPIHT we obtained the reconstructed Lena image is shown in fig.4. (c). With PSNR is42.845dB and MSE is 3.3772.Some of the best known algorithms in the literature. Comparing with improvised JPEG, DTT-SPIHT shows a high PSNR on Lena image. Advantage of the DTT-SPIHT method is its simplicity in comparison to improved JPEG. Entropy encoding and decoding processes are no DTT-SPIHT shows a good PSNR gain over existing image compression schemes on Lena image. DTT- SPIHT always performs the better than DCT-SPIHT scheme for Lena image.

    The performance DTT-SPIHT algorithm shows a high PSNR compared with DCT-SPIHT.

    Where, T(I,j) is the coefficient at node (I,j).T is the testing coordinate set..Briefly, SPIHT coding method is described in the following two passes:

    Initialization: LSP contains no elements (empty). LIP contains all tree roots at coarsest scale.LIS contains all tree nodes; choose the threshold according to Sn(T) .

    • Sorting pass: output Sn(I,j) for all the nodes (I,j) of LIP. If Sn=1, move the node (I,j) to the LSP and output the sign of T (I,j). for all the nodes belongs to type A. output Sn(D(I,j)). If Sn (D(I,j))=1, output Sn(K,L) for any node (k,l)


      If sn(k,l)=1, append node (k,l) to the LSP and output its sign. Otherwise, move (k,l) to the LIP. If L (I,j), then move (I,j) from LIS,

      Fig.4. (a) Input Image

      Fig.4. (b).Compressed Image

      Fig.4. (c).Reconstructed Image

  6. Conclusion

    DTT-SPIHT image coding scheme is implemented. It has been demonstrated that the implemented image coding algorithm shows an impressive PSNR gain over standard baseline Image compression standards on Lena image. For image having sharp edges, DTT- SPIHT consistently performs better than DCT-SPIHT. Future research direction is to use adaptive HVS and modified SPIHT algorithm, which is expected to improve the image quality at low bit-rates both subjectively and objectively

  7. References

  1. C. Chrysie_ s and A. Ortega, \Line Based Reduced Memory Wavelet Image Compression," in Proc. IEEE Data Compression Conference, (Snowbird, Utah), pp. 398{407, 1998.

  2. W. Penne baker and J. Mitchell, JPEG Still Image Data Compression Standard. Van No strand Reinhold, 1994.

  3. D. Lee, \New work item proposal: JPEG2000 image coding system." ISO/IEC JTC1/SC29/WG1 N390, 1996.

  4. J. M. Shapiro, \Embedded Image Coding Using Zerotrees ofWavelet Coe_cients," IEEE Trans. Signal Processing, vol. 41, pp. 3445{3462, December 1993.

  5. habibollah danyoli and Alfred mertins .Highly scalable image compression based on spiht for network applications

  6. Rafael C.Gonzalez,Richard E.Woods,Digital Image Processing,Prentice-Hall,2002 .

  7. Rafael C.Gonzalez, Richard E. Woods,Steven l.Eddins ,Digital Image Processing Using MATLAB,Prentice-Hall, 2004.

  8. A. M. Tekalp, Digital Video Processing. Englewood Cliffs, NJ: Prentice-Hall, 1995.

  9. M. Antonini, M. Barlaud, P. Mathieu, and I. Daubechies, Image coding using wavelet transform, IEEE Trans. Image Processing, vol. 1, pp. 205220, Apr. 1992.

  10. D. Taubman and A. Zakhor, Multirate 3-D subband coding of video, IEEE Trans. Image Processing, vol. 3, pp. 572588, Sept. 1994.

Leave a Reply