Comparative Analysis on Medical Images using SPIHT, STW and EZW

Comparative Analysis on Medical Images using SPIHT, STW and EZW

Jayant Kumar Rai Mr.Chandrashekhar Kamargaonkar

ME (Communication) Student Associate Professor, Dept. of ET&T FET-SSGI, SSTC, BHILAI FET-SSGI, SSTC, BHILAI

Chhattisgarh, INDIA Chhattisgarh, INDIA

Abstract Pathology labs are produces large amount of digital images. Digital images are required considerably large memory. However the only solution is a lossless compression technique. In this paper we compared the wavelet based algorithms such are Embedded Zero tree Wavelet (EZW), Set partitioning in Hierarchical Trees (SPIHT) and spatial orientation tree Wavelet (STW). These algorithms are compared in terms of Compression Ratio (CR), Peak Signal to Noise Ratio (PSNR), Bit per pixels and Mean Square Error (MSE). STW is the best algorithms as compared to others image compression method. It gives high CR and PSNR with better image quality.

Keywords Compression Ratio, EZW, Image Compression, MSE, SPIHT and STW.

1. INTRODUCTION

   Image compression is combined with two words image and compression. Image is refers to a two dimensional array of finite size, finite number of bits represent the digital image.

   Image are creating from digital camera requires large amount of storage capacity and bandwidth. To reduce the storage capacity and bandwidth we need an image compression. Compression is refers to reducing the quality of information used to represent a file, image or video without reducing the original information. Lossless and Lossy are the two parts of image compression. In lossless compression technique represents the original image but it gives low compression ratio. Lossless compression techniques are JPEGLS, Discrete Wavelet Transformed and Set Partitioning In Hierarchical Trees. In Lossy compression technique some amount of information is lost but it gives high compression ratio.LossycompressiontechniqueisDiscreteCosinetransforme dandDiscreteWaveletTransformed [1][2][4][5].Image Compression are used to reduce the redundancy present inside the image such as inter pixel , coding and psycho visual redundancy. Wavelet transformed is an important tool for image processing. It divides the image into number of segments each corresponding to a different frequency band. Discrete wavelet transform and Continuous wavelet transformed are the classification of wavelet transform. Discrete Wavelet transformed are Embedded Zero tree Wavelet (EZW), Set partitioning in Hierarchical Trees (SPIHT) and spatial orientation tree Wavelet (STW). DWT are widely used in image processing because it localize the signal in both Time and frequency domain. Application of Discrete wavelet transformed are compression, Transformation, De-noising and characterization [1][2]. In this paper we compared the wavelet based algorithms such

   As Embedded Zero tree Wavelet (EZW), Set partitioning in Hierarchical Trees (SPIHT) and Spatial orientation tree Wavelet (STW). These algorithms are compared in terms of Compression Ratio (CR), Peak Signal to Noise Ratio (PSNR), Bit per pixels and Mean Square Error (MSE).Our paper is divided into 7 sections such as Section II- SPIHT, section III-deal with STW, section IV -EZW ,section V- performance parameters section VI -deal result and analysis, section VII-conclusion followed by References.

2. SET PARTITIONING IN HIERARCHICAL TREES (SPIHT)

   Fig.1 SPIHT

   Set partitioning in Hierarchical Trees (SPIHT) is the most important algorithms based on wavelet coefficient. In SPIHT technique it converts the image coefficient into wavelet coefficient at the encoder side and the decoder performed the inverse operation i.e. it convert the wavelet coefficient into image coefficient [1][2].

   Characteristics of SPIHT are as following:-

   1. It performed fast encoding and decoding.

   2. It is a lossless compression technique.

   3. It gives high Peak signal to noise ratio and compression ratio.

   4. It produce the image with better quality.

   In SPIHT technique, it divide the wavelet into Spatial orientation trees. Each node in a tree has a individual pixels. In SPIHT method the information about the most significant bits of the wavelet element will be transmitted first on lower order significant bits. In each bit plane inforamtion is transmitted in three bits:-

   1. List of Insignificant Sets (LIS)

   2. List of Significant pixels(LIP)

   3. List of insignificant pixels(LSP)

   The pixels values are compared to certain threshold level.If pixels values are greater then threshold is called LSP. If pixels values are lower then threshold is called LIP. If pixels values are approximatelyt equal to hreshold is called LIS[1][2].

   Iitialize : All Roots Initialize : Empty Initialize : Ds Roots

   EZW is used adaptive arithmetic coding. It is used to achieve a fast and efficient method for entropy coding. The most important element of EZW algorithm is encoding process [11][12].

   LIP

   (I,j)

   LIP

   (I,j)

   If significant

   LSP

   (l,k)

   LSP

   (l,k)

   if significant

   LIS

   D(m,n) or L(m,n)

   Insignificant

   Significant

   Fig 3 Spatial Orientation Trees

   Single Element Subsets

   os stripped from Ds

   Set Partitioning

   Operation

   Multiple Element Subsets Remainder from Ds as Ls or

   Os(as Ds) stripped from Ls

   V. PERFORMANCE PARAMETERS

   In all images following performance parameters are used:-

   1. MEAN SQURE ERROR (MSE):-

      MSE= 1 1 1[(, ) (, )]2——— (1)

      Fig2 SPIHT Shorting Pass

      =

      =

3. SPATIAL ORIENTATION TREE WAVELET (STW)

   STW is based on state transition modal. From one threshold to the next the locations of transformed values undergo state transitions. State transitions modal allow the STW to reduce the number of bits needed for encoding. The difference between SPIHT and STW is that SPIHT is more careful in its organization of coding output. The difference between STW and EZW is that STW uses a different approach to encoding the zero tree information [11][12].

   Where Io(x, y)= original image

   Ic(x, y)= compressed image M = dimensions of the images.

   In the image error will be low when the value of MSE is low.

   1. Mean Absolute Difference (MAD):-

      MAD = 1 1 1[(, ) (, )]. ———– (2)

      =

      =

4. EMBEDDED ZERO TREE WAVELET (EZW)

1. PEAK SIGNAL TO NOISE RATIO (PSNR):-

Embedded Zero Tree Wavelet is the simple algorithm used for image compression. The EZW is based on the following principles:-

PSNR = 10 log10

(255255) —————— (3)

1. Prediction of the absence of significancent information across scales.

2. Entropy coded successive approximation quantization and universal lossless information compression which is accomplised through arithmetic

From equation (3) we observed that PEAK SIGNAL TO

NOISE RATIO (PSNR) is inversely proportional to the MEAN SQURE ERROR (MSE) i.e. higher value of PSNR can be achieved by decreasing the value of MSE.

4) COMPRESSION RATIO (CR):-

coding.Discrete wavelet transferred or hierarchical subband decomposition .

CR = 0

——————– (4)

Where = original image size

= compress image size.

1. RESULT AND ANALYSIS

   Medical images has taken from the Image diagnostic centre BHILAI . This experment is performed on medical images by using MATLAB (2012 a).The dimensional size of MRI images are 512 x 512. This experiment is performed on more than 300 medical images by using Embedded Zero tree Wavelet (EZW), Set partitioning in Hierarchical Trees (SPIHT) and Spatial orientation tree Wavelet (STW).

   Fig 4 Image 1

   Fig 7 Image 4

   Image Sequence	Compression Ratio
   	SPIHT	STW	EZW
   1	7.29	10.98	11.15
   2	7.21	10.85	10.09
   3	4.76	7.18	7.17
   4	4.66	7.03	7.13
   5	4.72	7.07	7.09
   AVERAGE	5.728	8.622	8.526

   Image Sequence	Compression Ratio
   	SPIHT	STW	EZW
   1	7.29	10.98	11.15
   2	7.21	10.85	10.09
   3	4.76	7.18	7.17
   4	4.66	7.03	7.13
   5	4.72	7.07	7.09
   AVERAGE	5.728	8.622	8.526

   Fig 8 Image 5 TABLE I

   Fig 5 Image 2

   12

   COMPRESSION RATIO

   COMPRESSION RATIO

   10

   8

   6

   4

   2

   0

   1 2 3 4 5

   IMAGE SEQUENCE

   SPIHT STW EZW

   Fig 6 Image 3

   Fig 9 COMPRESSION RATIO OF MRI IMAGE

   Image Sequence	Peak Signal to Noise Ratio
   	SPIHT	STW	EZW
   1	39.58	41.24	40.37
   2	39.06	41.13	39.74
   3	40.79	42.3	41.42
   4	41.07	42.32	41.74
   5	40.93	42.29	41.57
   AVERAGE	40.286	41.856	40.968

   Image Sequence	Peak Signal to Noise Ratio
   	SPIHT	STW	EZW
   1	39.58	41.24	40.37
   2	39.06	41.13	39.74
   3	40.79	42.3	41.42
   4	41.07	42.32	41.74
   5	40.93	42.29	41.57
   AVERAGE	40.286	41.856	40.968

   TABLE II

   TABLE IV

   Image Sequence	Bit per Pixels
   	SPIHT	STW	EZW
   1	0.584	0.879	0.892
   2	0.872	0.868	0.577
   3	0.38	0.575	0.574
   4	0.37	0.563	0.571
   5	0.388	0.566	0.567
   AVERAGE	0.5188	0.6902	0.6362

   43

   42

   PSNR

   PSNR

   41

   40

   39

   38

   37

   1 2 3 4 5

   IMAGE SEQUENCE

   SPIHT STW EZW

   1

   0.9

   BIT PER PIXELS

   BIT PER PIXELS

   0.8

   0.7

   0.6

   0.5

   0.4

   0.3

   0.2

   0.1

   0

   1 2 3 4 5

   IMAGE SEQUENCE

   SPIHT STW EZW

   Fig 10 PSNR OF MRI IMAGE

   Image Sequence	Mean Square Ratio
   	SPIHT	STW	EZW
   1	7.14	4.87	5.95
   2	8.06	5	6.9
   3	5.41	3.82	4.67
   4	5.07	3.81	4.34
   5	5.24	3.83	4.52
   AVERAGE	6.184	4.266	5.276

   Image Sequence	Mean Square Ratio
   	SPIHT	STW	EZW
   1	7.14	4.87	5.95
   2	8.06	5	6.9
   3	5.41	3.82	4.67
   4	5.07	3.81	4.34
   5	5.24	3.83	4.52
   AVERAGE	6.184	4.266	5.276

   TABLE III

   Fig 12 BPP OF MRI IMAGE

   From Table I and Figure 9 it is easy identify that STW has better Compression ratio as compared to SPIHT and EZW. Average compression ratio of STW, SPIHT and EZW are 8.62, 5.72 and 8.52 respectively. From Table II and Figure 10 STW gives high PSNR as compared to SPIHT and EZW. Average PSNR of STW, SPIHT and EZW are 41.856, 6.184 and 40.968 respectively. From Table III and Figure 11 STW gives low MSE as compared to SPIHT and EZW. Average MSE of STW, SPIHT and EZW are 4.266, 40.286 and 5.276

   respectively.

   ERROR

   ERROR

   9

   8

   7

   SQUARE

   SQUARE

   6

   5

   4

   3

   MEAN

   MEAN

   2

   1

   0

   1 2 3 4 5

   IMAGE SEQUENCE

   Fig 11MSE OF MRI IMAGE

   SPIHT STW EZW

2. CONCLUSION

In this paper, we compared the Embedded Zero tree Wavelet (EZW), Set partitioning in Hierarchical Trees (SPIHT) and spatial orientation tree Wavelet (STW). From Result and analysis we say that STW gives high CR and PSNR with better image quality.

REFERENCES

1. M.Fermi ukrit and G.R. Suresh " Motion Estimation and Motion Compensated Medical Image Sequence Compression Using SPIHT " International conference on Advanced Computing (ICoAC

   ),India,IEEE,2013.

2. Shaou-Gang Miaou,Member IEEE,Fu-sheng ke and Shu-ching chen "A Lossless compression method for Medical Image Sequence using JPEG- LS and Interframe coding"IEEE Transaction on Information Technology in Biomedicine ,Volume 13,No. 5,sep 2009

3. Thazni Aziz,D.Raveena Judie Dolly " Motion Estimation and Motion Compensated video compression using DCT and DWT ",International Journal of Emerging Technology and Advanced Engineering ,VOL 2, issue 12, DEC 2012..

4. Shan-Zhu and Kai-Kuang Ma "A New Diamond Search Algorithm for Fast Block Matching Motion Estimation " ,IEEE Transactions on image processing , vol 9, No 2, feb 2000.

5. T.M.P Rajkumar and Mrityunjaya V.Latte "Novel Superimposed Diamond Search Algorithm for medical image compression " International Journal Computer Applications in Technology ,vol 51 ,No 3, 2015.

6. Aroh Barjatya ,student member , IEEE " Block Matching Algorithm For Motion Estimation " DIP 6620 SPRING 2004.

7. Arvind Kourav,Dr.Ashutosh Sharma " Comparative Analysis of Wavelet Transform Algorithm for Image Compression"InternationalConference on Communication and Signal Processing , April 3-5 ,2014 , India.

8. Eric chan,student Member ,IEEE and Sethuraman Panchanathan

   ,member , IEEE "Review of Block Matching Based Motion Estimation Algorithm for video compression "1993 IEEE.

9. YaoNie,student member,IEEE and Kai-kuang Ma,senior member,IEEE Adaptive Rood Pattern Search for Fast Block Matching Motion Estimation IEEE Transactions on Iamge processing, VOL 11,No.12,December 2002.

10. Chandandeep Kaur,Sumit Budhiraja, Improvement of SPIHT in Image Compression Survey , International Journal of Emerging Technology and Advanced Engineering ,vol.03,Issue 1,January 2013.

11. Rohit kumar , Dr. Sukhwinder Singh Image Compression using STW and WDR wavelet IJIRST,Volume1,Issue3,August 2014.

12. Arvind kourav, Dr. ashutosh Sharma Comparative Analysis of wavelet transform algorithms for image compression Internation Conference on communication and signal processing, April 3-5,2014.