Anda di halaman 1dari 20

Hindawi Publishing Corporation EURASIP Journal on Applied Signal Processing Volume 2006, Article ID 92734, Pages 113 DOI

10.1155/ASP/2006/92734

A Comparison of Set Redundancy Compression Techniques


Samy Ait-Aoudia and Abdelhalim Gabis
Institut National dInformatique (INI), BP 68M, Oued Smar 16270, Algiers, Algeria Received 27 February 2005; Revised 30 November 2005; Accepted 21 January 2006 Medical imaging applications produce large sets of similar images. Thus a compression technique is necessary to reduce space storage. Lossless compression methods are necessary in such critical applications. Set redundancy compression (SRC) methods exploit the interimage redundancy and achieve better results than individual image compression techniques when applied to sets of similar images. In this paper, we make a comparative study of SRC methods on sample datasets using various archivers. We also propose a new SRC method and compare it to existing SRC techniques. Copyright 2006 Hindawi Publishing Corporation. All rights reserved.

1.

INTRODUCTION

Medical imaging applications produce a huge amount of similar images. Storing such amount of data needs gigantic disk space. Thus a compression technique is necessary to reduce space storage. In addition, medical images must be stored without any loss of information since the delity of images is critical in diagnosis. This requires lossless compression techniques. Lossless compression is an error-free compression. The decompressed image is the same as the original image. Classical image compression techniques (see [15]) concentrate on how to reduce the redundancies presented in an individual image. These compression techniques use the same model of compression as shown in Figure 1. This model ignores an additional type of redundancy that exists in sets of similar images, the set redundancy. The term set redundancy was introduced for the rst time by Karadimitriou [6] and dened as follows: Set redundancy is the interimage redundancy that exists in a set of similar images, and refers to the common information found in more than one image in the set. The compression techniques based on set redundancy follow the model presented in Figure 2. These methods are referred to as SRC (for set redundancy compression) methods. After extracting the set redundancy, any compression algorithm can be applied to achieve higher compression ratios. In this paper, we present an evaluation of the set redundancy compression (SRC) methods combined with dierent archivers. The SRC methods tested are the Min-Max differential method (MMD), the Min-Max predictive (MMP) method, and centroid method. The archivers used for individual compression are RAR compressor which is based on

[79], Gzip which is a variation of Ziv-Lempel (1977) [9] method, Bzip2 that uses Ziv-Lempel (1978) [10] method, and the ZIP archiver. The Human encoder [7] is also used in the evaluation. This paper is organized as follows. We dene, in Section 2, the correlation coecient to quantify similarity between images. The dierent SRC methods are explained in Section 3. We present, in Section 4, a new predicting scheme for the Min-Max predictive method. Experimental results on medical CT (computed tomography) and MR (magnetic resonance) brain images are given in Section 5. Section 6 gives conclusions. 2. IMAGES SIMILARITY

The redundancy extraction is a worth operation if the images in the set are similar. The visual impression is not sucient to state that two or more images are similar. We must have a statistical criterion to test similarity. Two images are said to be similar or statistically correlated if they have similar pixel intensities in the same areas or they have comparable histograms. The correlation coecient is used to quantify similarity. For two datasets X = (x1 , x2 , . . . , xN ) and Y = ( y1 , y2 , . . . , yN ) with mean values xm and ym , Neter et al. [11] dened this coecient as r=
N i=1 N i=1

xi xm y i y m
2 N i=1

xi xm

yi ym

(1)

The correlation coecient is also called Persons r . To avoid the manipulation of negative values, r 2 is often used

2
Original image Individual image compression Compressed image

EURASIP Journal on Applied Signal Processing (2) decoder:


Di + mini

Figure 1: Standard compression model.

value Pi =

maxi Di

if value Pi mini < maxi value Pi ; otherwise,

(3)

instead of r . For two datasets X and Y , a value of r 2 close to 0 means that no correlation exists between them. A value of r 2 close to 1 means that strong correlation exists between the two datasets. X and Y are perfectly correlated if r 2 = 1. In context of images, a value r 2 close to 0 means that the two images are totally dissimilar, a value r 2 close to 1 indicates strong similarity, and a value r 2 = 1 means that the images are identical. We give two examples to test the existence of correlation among images. Figure 3 shows two successive MRI brain scans of the same patient. The value r 2 = 0.80 indicates strong similarity between these two images. Figure 4 depicts two nonsimilar images. The correlation parameter r 2 = 0.005 indicates that the two images are noncorrelated. 3. SET REDUNDANCY METHODS

where Di , is the dierence value to be stored in the dierence image, mini is the value at position i in the MIN image and maxi is the value at position i in the MAX image. To synchronize encoding and decoding, the encoder uses consistently Min or Max curves until it nds a dierence value larger than (max min)/ 2. In that case, it encodes this value and switches to the other curve. The decoder follows the same rule; when it nds a dierence larger than (max min)/ 2, it also switches to the other curve. 3.2. Min-Max predictive method

In this section we present four types of SRC methods: the Min-Max dierential method [6, 12], the Min-Max predictive method [6, 13], the centroid method [6, 14], and the multilevel centroid method [15]. These methods are fast, lossless, and easy to implement. 3.1. Min-Max differential method MMD uses, for extracting the set redundancy in a set of similar images, two images: a maximum image and a minimum image. To create the minimum (MIN) image, the pixel values across all the images in the set are compared, and for each pixel position the smallest value is chosen. Similarly, the maximum (MAX) image is created by selecting the largest pixel value for each pixel position. Then, the set redundancy can be reduced by replacing every image in the set by its dierences from the min or the max image, such that for every pixel position, MMD nds and stores the smallest dierence value (see Figure 5). Note that pixel values are indexed with only one subscript, despite corresponding to a two-dimensional array. The image is observed pixel by pixel in a predened raster scan order. The algorithms of both encoder and decoder are presented below. For each pixel at position i: (1) encoder:
value Pi mini

The MMP method also uses the Min and Max images. It is more elaborated than the MMD method but it is also a more powerful method. For each pixel at position i, the MIN image provides the minimal value mini of all the images, and the image MAX provides the maximum value maxi . These two values are the limits of the range of the possible values that a pixel at position i can have in each image in the set. After dividing this interval into N levels, a pixel at position i in each image can be represented as a level Li between its corresponding minimum and maximum values (see Figure 6). The level Li is given by the equation Value Pi mini , maxi mini

Li = N

(4)

where Li is the level of a pixel at position i in a given image, and N is number of levels (N = 256). Neighboring pixels often have similar levels despite having dierent values. For example, consider the values of the following neighboring pixels given in Table 1. From (4), a prediction scheme for the value of pixel Pi can be dened as value predicted Pi = mini + Li maxi mini , N (5)

if value Pi mini < maxi value Pi , otherwise; (2)

Di =

maxi value Pi

where Li is the level predicted for a pixel at position i. The prediction concerns only the element Li in the preceding formula. The MMP method predicts the value of a pixel Pi by using the level information from already treated neighboring pixels. Since the levels of neighboring pixels are often similar, this is a good prediction scheme. Karadimitriou [6, 13] dened three predictors. These predictors determine three variations of Min-Max predictive methods referred to as MMP1, MMP2, and MMP3. The predictions schemes for MMP methods are shown in Table 2.

S. Ait-Aoudia and A. Gabis


Original image Set redundancy extraction Individual image compression (any method) Compressed image

Figure 2: Enhanced compression model.

255

Dierence values

Max image

Pixel values

Image from set

Min image (a) (b) 0 P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 Pixel positions

Figure 3: Two successive MRI brain scans.

Figure 5: Min-Max dierential method.

255

Levels

Max image

Pixel values

Image from set

Min image (a) (b) 0 P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 Pixel positions

Figure 4: Two dissimilar images. Figure 6: Min-Max predictive method (20 levels).

Lupper is the level of the upper neighboring pixel, Lleft is the level of the left neighbor, and Lupperleft is the level of the upper left neighbor (see Figure 7). For every image in the set, the encoding process consists of storing the dierences between the predicted values and the original values. These dierences values replace the original values. To restore the original image from the differences stored, the decoding process calculate the predicted values, and then adds the corresponding dierences values. 3.3. Centroid method The centroid method [6, 14] (which is also used in [16]), uses the average image of a set of similar images to predict the values of the dierence image. If the prediction is ecient enough, the dierence image will contain small values having a Laplacian distribution with most of values very close to zero.

A simple scheme for predicting the pixel value at position i in image j is Fi, j = mi , (6)

where mi is the average value at position i across all images and Fi, j is the predicted value. This scheme is not very ecient. A more sophisticated scheme [14] can be expressed as follows: Fi+1, j = mi+1 + xi, j mi , Di+1, j = xi+1, j Fi+1, j , (7)

where Fi+1, j is the predicted value at position i + 1, Xi, j is the pixel value at position i, mi is the average value of position i across all images, and Di+1, j is the dierence value of position i +1 in image j between the original and the predicted values. The detailed demonstration of (7) can be found in [6].

4
Table 1: Example of neighboring pixels levels. Pixel value 99 105 112 102 Min value 15 21 29 19 Maximum value 197 205 210 199 Level 118 117 117 118

EURASIP Journal on Applied Signal Processing

Pupperleft Pleft

Pupper Pi Current pixel

Figure 7: Notation used for specifying neighboring pixels of current pixel Pi.

Table 2: Level prediction in MMP methods. MMP method MMP1 MMP2 MMP3 Level prediction Li = Lleft Li = (Lupper + Lleft )/ 2 Li = Lupper + Lleft Lupperleft
e

c b

a x

d Current pixel

Figure 8: Notation used for specifying neighboring pixels of current pixel x.

3.4. Multilevel centroid method Proposed by El-Sonbaty et al. [15] and derived from the centroid method, this model executes the centroid method N levels times. Given a set of similar images X , the corresponding median image (median 1) is calculated. Applying the centroid method on the given input set, the dierence 1 set (dierence images at level 1) is obtained. Repeating the process recursively, the median 2 is obtained from the difference 1 set and applying centroid method again, the dierence 2 set is also obtained. The process stops when all levels are processed. The rst level is the centroid method. The prediction scheme of this method is the same as the centroid method, and is given by Fi+1, j (n) = mi+1 (n) + xi, j (n) mi (n), Di+1, j (n) = xi+1, j (n) Fi+1, j (n),

not detected, then the guessed value is a + b c. Specically, the LOCO-I predictor guesses
min(a, b) a + b c

if c max(a, b), (9)

predicted x = max(a, b) if c min(a, b), otherwise.

(8)

LOCO-I is the algorithm at the core of the ISO/ITU/ 14495-1 standard for compression of continuous-tone images, JPEG-LS (see [18]). The guessed value is seen as the median of three xed predictors a, b, and a+b c. The predictor used in LOCO-I was renamed during the standardization process median edge detector (MED). From the MED predictor we derive a new predicting scheme. In (5), the predicted term Li will be calculated as follows:
min Lupper , Lleft max Lupper , Lleft

if Lupperleft max Lupper , Lleft , if Lupperleft min Lupper , Lleft ,

where Fi+1, j (n) is the estimation of a pixel at position i + 1 in an image j at level n, xi, j (n) is the value of pixel i of the image j at level n, mi (n) is the value of pixel i of the median image at level n, and Di+1, j (n) is the value of pixel i of the dierence image j at level n. 4. THE NEW MMP PREDICTIVE SCHEME

Li = Lupper + Lleft

Lupperleft

otherwise, (10)

The three predictors used by Karadimitriou [6, 13] by assigning to Li (see Section 3.2) information from previous treated pixels are not exible. We propose to use a more elaborated predicting scheme. This scheme is based on the predictor used in Weinberger et al. proposal, LOCO-I (low complexity lossless compression for Images) [17]. LOCO-I uses a nonlinear predictor with edge detecting capability. It guesses the value of the current pixel x based on neighboring pixels (see Figure 8). The approach in LOCO-I consists in performing a primitive test to detect vertical or horizontal edges. If an edge is

where Lupper is the level of the upper neighboring pixel, Lleft is the level of the left neighbor, and Lupperleft is the level of the upper left neighbor. Since the image is processed pixel by pixel in a raster scan order, pixels of the rst line do not have upper left or upper neighbors. In this case, the value Lleft will be assigned to Li . Similarly, the value Lupper will be assigned to Li for pixels of the rst column in the image. Note that for the rst pixel of every image (no processed pixels yet), the value 128 is chosen to be the predicted level. The idea behind the use of the new predictor is to expect better results than those obtained by using predictors dened

S. Ait-Aoudia and A. Gabis in Section 3.2. We call the new method resulting from this predicting scheme MMPM for MMP MED. 5. EXPERIMENTAL RESULTS

The evaluation of set redundancy method is made on sample medical images. The images were taken from M.D. Anderson Cancer Center in Houston, Texas and Harvard Medical School. All images were gray-level, and were scaled to 8 bits/pixel. All experiments were performed under Windows XP operating system. To make the evaluation of the SRC methods, we have used the standard compression algorithms RAR, Bzip2, Gzip, ZIP, Human. The medical images are compressed by these algorithms with and without using the set redundancy extraction. Each algorithm is tested separately and the attained compression ratios are compared. The compression ratio is given by R= Size original image . Size compressed image (11)

The improvement against standard compression method is also needed in the evaluation. It shows if the use of SRC methods is really eective. The improvement in compression is dened by A= RSRC R , R (12)

where R is the compression ratio achieved when using a standard compression method only, and RSRC is the compression ratio achieved when combining SRC with that standard compression method. 5.1. M.D. Anderson Cancer Center images From M.D. Anderson Cancer Center images, a set of 10 CT (computed tomography) similar images, and another set of 10 MR images are chosen to conduct the rst tests. These two sets were selected and used by Karadimitriou [6, 1214] and also used by Sonbaty et al. [15], so an easy comparison can be made. The resolution is 512 512 for the CT images and 256 256 for the MR images. 5.1.1. CT experiments The sample set of computed tomography images used in the experiments is shown in Figure 9. The set contains axial CT brain scans where horizontal slices of the brain at the eyelevel are depicted. The images were selected from patients of both sexes, various ages, and a variety of pathological conditions. From the chosen set, the average, minimum, and maximum images were created to be used in the MMD, MMP, and centroid methods. These three images are shown in Figure 10. Results of tests on CT images (compression ratios and improvement in compression by using SRC methods) are presented in Table 3. The histograms representing
Figure 9: CT test images.

(a) Average CT image.

(b) Minimum CT image.

(c) Maximum CT image.

Figure 10: CT average, minimum, and maximum images.

EURASIP Journal on Applied Signal Processing


Table 3: Experimental results on CT images. Compression technique Original image Bzip2 Centroid + Bzip2 MMD + Bzip2 MMP1 + Bzip2 MMP2 + Bzip2 MMP3 + Bzip2 MMPM + Bzip2 Mutlilevel centroid (2 levels) + Bzip2 Gzip Centroid + Gzip Gzip + MMD MMP1 + Gzip MMP2 + Gzip MMP3 + Gzip MMPM + Gzip Mutlilevel centroid (2 levels) + Gzip Human Centroid + Human MMD + Human MMP1 + Human MMP2 + Human MMP3 + Human MMPM + Human Mutlilevel centroid (2 levels) + Human RAR Centroid + RAR MMD + RAR MMP1 + RAR MMP2 + RAR MMP3 + RAR MMPM + RAR Mutlilevel centroid (2 levels) + RAR ZIP Centroid + ZIP MMD + ZIP MMP1 + ZIP MMP2 + ZIP MMP3 + ZIP MMPM + ZIP Mutlilevel centroid (2 levels) + ZIP Average size (KO) 256 74.35 72.55 75.78 71.71 64.64 63.35 61.92 83.32 100.46 82.48 88.71 78.17 70.92 69.08 67.64 89.82 193.45 98.41 125.93 84.08 75.35 69.15 69.06 91.31 76.09 72.60 82.52 67.37 62.73 57.37 56.75 82.57 99.94 80.47 87.35 75.94 68.36 66.36 64.90 88.16 Average compression ratio 3.44 : 1 3.52 : 1 3.37 : 1 3.57 : 1 3.96 : 1 4.02 : 1 4.13 : 1 3.07 : 1 2.54 : 1 3.10 : 1 2.88 : 1 3.27 : 1 3.61 : 1 3.70 : 1 3.78 : 1 2.85 : 1 1.32 : 1 2.60 : 1 2.03 : 1 3.04 : 1 3.39 : 1 3.70 : 1 3.71 : 1 2.80 : 1 3.36 : 1 3.52 : 1 3.10 : 1 3.8 : 1 4.08 : 1 4.46 : 1 4.51 : 1 3.10 : 1 2.56 : 1 3.18 : 1 2.93 : 1 3.37 : 1 3.74 : 1 3.85 : 1 3.94 : 1 2.90 : 1 Improvement % 2
2

3 15 17 20
10

22 13 28 42 45 49 12 96 54 130 156 180 181 112 4


7

13 21 32 34
7

24 14 31 46 50 54 13

S. Ait-Aoudia and A. Gabis


200 150 100 50 0 Bzip2
50

Gzip

Human

RAR

Zip

MMD MMP1 MMP2 MMP3

MMPM Centroid Multilevel

Figure 11: SRC methods improvement on CT images.

5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 Bzip2 Gzip

Compression ratio

Human

RAR

Zip

Without SRC MMD MMP1 MMP2

MMP3 MMPM Centroid Multilevel

Figure 12: Average compression ratios on CT images.

improvements and compression ratios using SRC methods are shown in Figures 11 and 12, respectively. 5.1.2. MR experiments The set of magnetic resonance images scans depict is horizontal slices about 7-8 cm from the top of the head. These images are shown in Figure 13. From this set, the average, minimum, and maximum images were created to be used in the MMD, MMP, and centroid methods. These three images are presented in Figure 14. Results of tests on MR images (compression ratios and improvement in compression by using SRC methods) are presented in Table 4. The histograms representing improvements and compression ratios using SRC methods are shown in Figures 15 and 16, respectively.

Figure 13: MR test images.

(a) Average MR image.

(b) Minimum MR image.

(c) Maximum MR image.

Figure 14: Average, minimum, and maximum MR brain images.

EURASIP Journal on Applied Signal Processing


Table 4: Experimental results on MR images. Compression technique Original image Bzip2 Centroid + Bzip2 MMD + Bzip2 MMP1 + Bzip2 MMP2 + Bzip2 MMP3 + Bzip2 MMPM + Bzip2 Mutlilevel centroid (2 levels) + Bzip2 Gzip Centroid + Gzip MMD + Gzip MMP1 + Gzip MMP2 + Gzip MMP3 + Gzip MMPM + Gzip Mutlilevel centroid (2 levels) + Gzip Human Centroid + Human MMD + Human MMP1 + Human MMP2 + Human MMP3 + Human MMPM + Human Mutlilevel centroid (2 levels) + Human RAR Centroid + RAR MMD + RAR MMP1 + RAR MMP2 + RAR MMP3 + RAR MMPM + RAR Mutlilevel centroid (2 levels) + RAR ZIP Centroid + ZIP MMD + ZIP MMP1 + ZIP MMP2 + ZIP MMP3 + ZIP MMPM + ZIP Mutlilevel centroid (2 levels) + ZIP Average size (KO) 64 38.25 37.93 33.13 33.90 31.56 33.59 31.70 41.69 46.19 41.05 35.31 35.03 33.11 35.02 33.10 44.09 55.67 44.02 37.12 35.34 32.67 35.17 32.48 47.48 38.22 36.78 32.10 31.94 30.52 31.65 29.89 40.52 46.34 41.21 35.34 35.11 33.21 35.13 33.20 44.25 Average compression ratio 1.67 : 1 1.68 : 1 1.93 : 1 1.88 : 1 2.03 : 1 1.90 : 1 2.01 : 1 1.53 : 1 1.39 : 1 1.55 : 1 1.81 : 1 1.83 : 1 1.93 : 1 1.82 : 1 1.93 : 1 1.45 : 1 1.14 : 1 1.45 : 1 1.72 : 1 1.81 : 1 1.95 : 1 1.81 : 1 1.97 : 1 1.34 : 1 1.67 : 1 1.74 : 1 1.99 : 1 2.00 : 1 2.09 : 1 2.02 : 1 2.14 : 1 1.67 : 1 1.38 : 1 1.55 : 1 1.81 : 1 1.82 : 1 1.92 : 1 1.82 : 1 1.93 : 1 1.44 : 1 Improvement % 0.5 15 12 21 13 20
8

11 30 31 39 32 39 5 27 50 58 71 58 72 17 4 19 20 25 21 28 0 12 31 32 39 32 40 5

S. Ait-Aoudia and A. Gabis


80 70 60 50 40 30 20 10 0 10 20

9 The compression ratios attained on this set by using SRC methods are presented in Table 5. The histogram representing these compression ratios is shown in Figure 18. 5.2.2. Brain tumor images The set, shown in Figure 19, contains 30 axial MR brain scans. These images were selected from an MR brain exam of a 73-year old right-handed man that sought medical attention because of a grand mal seizure and progressive diculty with speech. The exam indicates the presence of a brain tumor. The compression ratios attained on this set by using SRC methods are presented in Table 6. The histogram representing these compression ratios is shown in Figure 20. 5.3. Discussion

Bzip2

Gzip

Human

RAR

Zip

MMD MMP1 MMP2 MMP3

MMPM Centroid Multilevel

Figure 15: SRC methods improvement on MR images.

2.5 Compression ratio 2 1.5 1 0.5 0 Bzip2 Without SRC MMD MMP1 MMP2 Gzip

Human MMP3 MMPM Centroid Multilevel

RAR

Zip

From the results shown in the previous tables on sample datasets, we see that the majority of SRC methods carry out an improvement compared to standard compression. This is a good indicator for the eectiveness of using SRC techniques on similar images datasets. The results show that, in most cases, the MMP methods perform better than the other SRC techniques. We also note that the proposed MMPM method attains compression ratios slightly better than the other MMP methods. The tests have also shown that the centroid and multilevel centroid techniques are not very ecient and that the Human encoder gives the worst compression ratios comparatively to other encoders when the number of images in the set grows. 6. CONCLUSION

Figure 16: Average compression ratios on MR images.

5.2. Harvard Medical School images From Harvard Medical School images, two sets of 20 and 30 magnetic resonance images are chosen to make the evaluation. These images are taken from the whole brain atlas which depicts various brain diseases. The resolution is 256 256 for all images. The images were converted to PGM format before being processed. 5.2.1. Cerebral edema images A sample set of medical images is shown in Figure 17. This set contains 20 axial MR brain scans. These images were selected from an MR brain exam of a 51-year old woman. The undertaken exam shows a cerebral edema which corresponds to the high signal extending from the center of the mass through surrounding white matter.

One of the best application areas for SRC methods is medical imaging. Medical image databases usually store huge amount of similar images (CT, MR, PET, Ultrasound, XRay, and Angiography images); therefore, they contain large amounts of set redundancy. This paper attempts to evaluate the performance of various SRC methods on sample datasets of grayscale similar images taken from dierent sources. An SRC method, called MMPM, is also proposed. It is based on the MED predictor of the JPEG-LS method. In the carried out tests, MMPM performs slightly better than the other MMP methods. We must mention that, to be eective, the SRC methods impose high similarity in the whole set of images. A preprocessing phase can be done to cluster similar images before launching the compression operation. In this study, only the eect of compressing sets of grayscale images was evaluated. Further works must consider compressing sets of multispectral or true color images. SRC methods can also be tested on many other application areas. Satellite image databases, for example, often contain sets of images taken over the same geographical areas, and under similar weather or lighting conditions. They necessarily contain interimage redundancy.

10

EURASIP Journal on Applied Signal Processing

Figure 17: MR brain scans.

Table 5: Average compression ratios on MR images. Without SRC Bzip2 Gzip Human RAR Zip 4.25 3.63 2.38 4.17 3.64 MMD 3.95 4.11 2.68 4.04 3.75 MMP1 4.02 4.30 2.86 4.20 3.99 MMP2 4.18 4.38 2.97 4.35 4.06 MMP3 4.08 4.35 2.81 4.27 4.01 MMPM 4.26 4.50 3.02 4.47 4.15 Centroid 3.53 3.45 2.40 3.60 3.40 Multilevel 3.38 3.33 2.36 3.40 3.32

S. Ait-Aoudia and A. Gabis


5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 Bzip2 Without SRC MMD MMP1 MMP2 Gzip

11

Compression ratio

Human MMP3 MMPM Centroid Multilevel

RAR

Zip

Figure 18: Average compression ratios on MR images.

Figure 19: MR brain scans.

12

EURASIP Journal on Applied Signal Processing


Table 6: Average compression ratios on MR images.

Bzip2 Gzip Human RAR Zip

Without SRC 5.37 4.97 2.99 4.98 4.97

MMD 5.12 5.12 3.14 5.02 5.05

MMP1 5.34 5.57 3.47 5.23 5.48

MMP2 5.51 5.66 3.55 5.52 5.60

MMP3 5.39 5.60 3.44 5.41 5.46

MMPM 5.56 5.78 3.56 5.60 5.68

Centroid 4.85 4.52 3.21 4.88 4.91

Multilevel 4.66 4.75 3.15 4.73 4.74

7 6 Compression ratio 5 4 3 2 1 0 Bzip2 Without SRC MMD MMP1 MMP2 Gzip Human MMP3 MMPM Centroid Multilevel RAR Zip

Figure 20: Average compression ratios on MR images.

ACKNOWLEDGMENT We would like to thank Kosmas Karadimiriou for the helpful discussions and Keith A. Johnson from Harvard Medical School for granting the use of the whole brain atlas medical images in the tests. REFERENCES
[1] H. Bekkouche and M. Barret, Adaptive multiresolution decomposition: application to lossless image compression, in IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 02), Orlando, Fla, USA, May 2002. [2] M. U. Celik, G. Sharma, and A. M. Tekalp, Gray-level embedded lossless image compression, in Proceedings of IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 03), pp. III-245III-248, Hong Kong, April 2003. [3] C. C. Chang and G. I. Chen, Enhancement algorithm for nonlinear context-based predictors, IEE Proceedings - Vision, Image, and Signal Processing, vol. 150, no. 1, pp. 1519, 2003. [4] D. A. Clunie, Lossless compression of grayscale medical images: eectiveness of traditional and state-of-the-art approaches, in Medical Imaging 2000: PACS Design and Evaluation: Engineering and Clinical Issues, vol. 3980 of Proceedings of SPIE, pp. 7484, San Diego, Calif, USA, February 2000. [5] J. Jiang, B. Guo, and S. Y. Yang, Revisiting the JPEG-LS prediction scheme, IEE Proceedings: Vision, Image and Signal Processing, vol. 147, no. 6, pp. 575580, 2000. [6] K. Karadimitriou, Set redundancy, the enhanced compression model, and methods for compressing sets of similar images, Ph.D. thesis, Department of Computer Science, Louisiana State University, Baton Rouge, La, USA, August 1996.

[7] D. A. Human, A method for the construction of minimum redundancy codes, Proceedings of IRE, vol. 40, no. 9, pp. 1098 1101, 1952. [8] D. Shkarin, Improving the eciency of PPM algorithm, Problems of Information Transmission, vol. 37, no. 3, pp. 226 235(10), 2001. [9] J. Ziv and A. Lempel, A universal algorithm for sequential data compression, IEEE Transactions on Information Theory, vol. 23, no. 3, pp. 337343, 1977. [10] J. Ziv and A. Lempel, Compression of individual sequences via variable-rate coding, IEEE Transactions on Information Theory, vol. 24, no. 5, pp. 530536, 1978. [11] J. Neter, W. Wasserman, and M. H. Kutner, Applied Linear Regression Models, IRWIN, Burr Ridge, Ill, USA, 1989. [12] K. Karadimitriou and J. M. Tyler, The min-max dierential method for large-scale storage and compression of medical images, in Proceedings of of Annual Molecular Biology and Biotechnology Conference, Baton Rouge, La, USA, 1996. [13] K. Karadimitriou and J. M. Tyler, Min-max compression methods for medical image databases, ACM SIGMOD Record, vol. 26, no. 1, pp. 4752, 1997. [14] K. Karadimitriou and J. M. Tyler, The Centroid method for compressing sets of similar images, Pattern Recognition Letters, vol. 19, no. 7, pp. 585593, 1998. [15] Y. El-Sonbaty, M. Hamza, and G. Basily, Compressing sets of similar medical images using multilevel centroid technique, in Processing of the 7th Conference on Digital Image Computing, Techniques and Applications, C. Sun, H. Talbot, S. Ourselin, and T. Adriaansen, Eds., Sydney, Australia, December 2003. [16] J. D. Lee, S. Y. Wan, and R. F. Wu, A hybrid compression model for clusters of similar medical images, Biomedical Engineering - Aplications, Basis & Communications, vol. 16, no. 1, 2003. [17] M. J. Weinberger, G. Seroussi, and G. Sapiro, LOCO-I: a low complexity, context-based, lossless image compression algorithm, in Proceedings of the IEEE Data Compression Conference, Snowbird, Utah, USA, April 1996, ISO Working Document ISO/IEC JTC1/SC29/WG1 N203. [18] M. J. Weinberger, G. Seroussi, and G. Sapiro, The LOCOI lossless image compression algorithm: principles and standardization into JPEG-LS, IEEE Transactions on Image Processing, vol. 9, no. 8, pp. 13091324, 2000. Samy Ait-Aoudia received a DEA Diplome dEtudes Approfondies in image processing from Saint-Etienne University, France, in 1990. He had a Ph.D. degree in computer science from Ecole des Mines, SaintEtienne, France, in 1994. He is currently Ma tre de Conf erences at the National Computer Science Institute in Algeria. He teaches dierent modules at both B.S. and M.S. levels in computer science and

S. Ait-Aoudia and A. Gabis


software engineering. His areas of research include image processing, CAD/CAM, and constraints management in solid modeling. Abdelhalim Gabis received the B.S. degree (Ing enieur dEtat en Informatique) from the National Computer Science Institute, Algiers, Algeria, in 2002. He had an M.S. degree from the same institute in 2005. He is currently working toward the Ph.D. degree at the National Computer Science Institute in Algeria. His research interests include data compression as well as image/video coding and processing. He is a Member of the Perceptions Research Group supported by the Research Ministry in Algeria.

13

EURASIP JOURNAL ON APPLIED SIGNAL PROCESSING

Special Issue on Transforming Signal Processing Applications into Parallel Implementations


Call for Papers
There is an increasing need to develop ecient systemlevel models, methods, and tools to support designers to quickly transform signal processing application specication to heterogeneous hardware and software architectures such as arrays of DSPs, heterogeneous platforms involving microprocessors, DSPs and FPGAs, and other evolving multiprocessor SoC architectures. Typically, the design process involves aspects of application and architecture modeling as well as transformations to translate the application models to architecture models for subsequent performance analysis and design space exploration. Accurate predictions are indispensable because next generation signal processing applications, for example, audio, video, and array signal processing impose high throughput, real-time and energy constraints that can no longer be served by a single DSP. There are a number of key issues in transforming application models into parallel implementations that are not addressed in current approaches. These are engineering the application specication, transforming application specication, or representation of the architecture specication as well as communication models such as data transfer and synchronization primitives in both models. The purpose of this call for papers is to address approaches that include application transformations in the performance, analysis, and design space exploration eorts when taking signal processing applications to concurrent and parallel implementations. The Guest Editors are soliciting contributions in joint application and architecture space exploration that outperform the current architecture-only design space exploration methods and tools. Topics of interest for this special issue include but are not limited to:
modeling applications in terms of (abstract) joint application and architecture performance

analysis
extending the concept of algorithmic engineering to

architecture engineering
design cases and applications mapped on

multiprocessor, homogeneous, or heterogeneous SOCs, showing joint optimization of application and architecture Authors should follow the EURASIP JASP manuscript format described at http://www.hindawi.com/journals/asp/. Prospective authors should submit an electronic copy of their complete manuscript through the EURASIP JASP manuscript tracking system at http://www.hindawi.com/mts/, according to the following timetable:
Manuscript Due Acceptance Notication Final Manuscript Due Publication Date September 1, 2006 January 1, 2007 April 1, 2007 3rd Quarter 2007

GUEST EDITORS: F. Deprettre, Leiden Embedded Research Center, Leiden University, Niels Bohrweg 1, 2333 CA Leiden, The Netherlands; edd@liacs.nl Roger Woods, School of Electrical and Electronic Engineering, Queens University of Belfast, Ashby Building, Stranmillis Road, Belfast, BT9 5AH, UK; r.woods@qub.ac.uk Ingrid Verbauwhede, Katholieke Universiteit Leuven, ESAT-COSIC, Kasteelpark Arenberg 10, 3001 Leuven, Belgium; Ingrid.verbauwhede@esat.kuleuven.be Erwin de Kock, Philips Research, High Tech Campus 31, 5656 AE Eindhoven, The Netherlands; erwin.de.kock@philips.com

control-dataow graph, dataow graph, and process network models of computation (MoC) transforming application models or algorithmic engineering transforming application MoCs to architecture MoCs joint application and architecture space exploration

Hindawi Publishing Corporation http://www.hindawi.com

EURASIP JOURNAL ON APPLIED SIGNAL PROCESSING

Special Issue on Video Adaptation for Heterogeneous Environments


Call for Papers
The explosive growth of compressed video streams and repositories accessible worldwide, the recent addition of new video-related standards such as H.264/AVC, MPEG-7, and MPEG-21, and the ever-increasing prevalence of heterogeneous, video-enabled terminals such as computer, TV, mobile phones, and personal digital assistants have escalated the need for ecient and eective techniques for adapting compressed videos to better suit the dierent capabilities, constraints, and requirements of various transmission networks, applications, and end users. For instance, Universal Multimedia Access (UMA) advocates the provision and adaptation of the same multimedia content for dierent networks, terminals, and user preferences. Video adaptation is an emerging eld that oers a rich body of knowledge and techniques for handling the huge variation of resource constraints (e.g., bandwidth, display capability, processing speed, and power consumption) and the large diversity of user tasks in pervasive media applications. Considerable amounts of research and development activities in industry and academia have been devoted to answering the many challenges in making better use of video content across systems and applications of various kinds. Video adaptation may apply to individual or multiple video streams and may call for dierent means depending on the objectives and requirements of adaptation. Transcoding, transmoding (cross-modality transcoding), scalable content representation, content abstraction and summarization are popular means for video adaptation. In addition, video content analysis and understanding, including low-level feature analysis and high-level semantics understanding, play an important role in video adaptation as essential video content can be better preserved. The aim of this special issue is to present state-of-theart developments in this ourishing and important research eld. Contributions in theoretical study, architecture design, performance analysis, complexity reduction, and real-world applications are all welcome. Topics of interest include (but are not limited to):
Heterogeneous video transcoding Scalable video coding Dynamic bitstream switching for video adaptation Signal, structural, and semantic-level video

adaptation
Content analysis and understanding for video

adaptation Video summarization and abstraction Copyright protection for video adaptation Crossmedia techniques for video adaptation Testing, eld trials, and applications of video adaptation services International standard activities for video adaptation

Authors should follow the EURASIP JASP manuscript format described at http://www.hindawi.com/journals/asp/. Prospective authors should submit an electronic copy of their complete manuscript through the EURASIP JASP manuscript tracking system at http://www.hindawi.com/mts/, according to the following timetable:
Manuscript Due Acceptance Notication Final Manuscript Due Publication Date September 1, 2006 January 1, 2007 April 1, 2007 3rd Quarter 2007

GUEST EDITORS: Chia-Wen Lin, Department of Computer Science and Information Engineering, National Chung Cheng University, Chiayi 621, Taiwan; cwlin@cs.ccu.edu.tw Yap-Peng Tan, School of Electrical and Electronic Engineering, Nanyang Technological University, Nanyang Avenue, Singapore 639798, Singapore; eyptan@ntu.edu.sg Ming-Ting Sun, Department of Electrical Engineering, University of Washington, Seattle, WA 98195, USA ; sun@ee.washington.edu Alex Kot, School of Electrical and Electronic Engineering, Nanyang Technological University, Nanyang Avenue, Singapore 639798, Singapore; eackot@ntu.edu.sg

Anthony Vetro, Mitsubishi Electric Research Laboratories, 201 Broadway, 8th Floor, Cambridge, MA 02138, USA; avetro@merl.com

Hindawi Publishing Corporation http://www.hindawi.com

EURASIP JOURNAL ON APPLIED SIGNAL PROCESSING

Special Issue on Knowledge-Assisted Media Analysis for Interactive Multimedia Applications


Call for Papers
It is broadly acknowledged that the development of enabling technologies for new forms of interactive multimedia services requires a targeted conuence of knowledge, semantics, and low-level media processing. The convergence of these areas is key to many applications including interactive TV, networked medical imaging, vision-based surveillance and multimedia visualization, navigation, search, and retrieval. The latter is a crucial application since the exponential growth of audiovisual data, along with the critical lack of tools to record the data in a well-structured form, is rendering useless vast portions of available content. To overcome this problem, there is need for technology that is able to produce accurate levels of abstraction in order to annotate and retrieve content using queries that are natural to humans. Such technology will help narrow the gap between low-level features or content descriptors that can be computed automatically, and the richness and subjectivity of semantics in user queries and high-level human interpretations of audiovisual media. This special issue focuses on truly integrative research targeting of what can be disparate disciplines including image processing, knowledge engineering, information retrieval, semantic, analysis, and articial intelligence. High-quality and novel contributions addressing theoretical and practical aspects are solicited. Specically, the following topics are of interest:

Authors should follow the EURASIP JASP manuscript format described at http://www.hindawi.com/journals/asp/. Prospective authors should submit an electronic copy of their complete manuscript through the EURASIP JASP manuscript tracking system at http://www.hindawi.com/mts/ according to the following timetable:

Manuscript Due Acceptance Notication Final Manuscript Due Publication Date

September 1, 2006 January 15, 2007 April 1, 2007 3rd Quarter, 2007

GUEST EDITORS: Ebroul Izquierdo, Department of Electronic Engineering, Queen Mary, University of London, Mile End Road, London E1 4NS, United Kingdom; ebroul.izquierdo@elec.qmul.ac.uk Hyoung Joong Kim, Department of Control and Instrumentation Engineering, Kangwon National University, 192 1 Hyoja2 Dong, Kangwon Do 200 701, Korea; khj@kangwon.ac.kr Thomas Sikora, Communication Systems Group, Technical University Berlin, Einstein Ufer 17, 10587 Berlin, Germany; sikora@nue.tu-berlin.de

Semantics-based multimedia analysis Context-based multimedia mining Intelligent exploitation of user relevance feedback Knowledge acquisition from multimedia contents Semantics based interaction with multimedia Integration of multimedia processing and Semantic Web technologies to enable automatic content sharing, processing, and interpretation Content, user, and network aware media engineering Multimodal techniques, high-dimensionality reduction, and low level feature fusion

Hindawi Publishing Corporation http://www.hindawi.com

EURASIP JOURNAL ON APPLIED SIGNAL PROCESSING

Special Issue on Super-resolution Enhancement of Digital Video


Call for Papers
When designing a system for image acquisition, there is generally a desire for high spatial resolution and a wide eldof-view. To achieve this, a camera system must typically employ small f-number optics. This produces an image with very high spatial-frequency bandwidth at the focal plane. To avoid aliasing caused by undersampling, the corresponding focal plane array (FPA) must be suciently dense. However, cost and fabrication complexities may make this impractical. More fundamentally, smaller detectors capture fewer photons, which can lead to potentially severe noise levels in the acquired imagery. Considering these factors, one may choose to accept a certain level of undersampling or to sacrice some optical resolution and/or eld-of-view. In image super-resolution (SR), postprocessing is used to obtain images with resolutions that go beyond the conventional limits of the uncompensated imaging system. In some systems, the primary limiting factor is the optical resolution of the image in the focal plane as dened by the cut-o frequency of the optics. We use the term optical SR to refer to SR methods that aim to create an image with valid spatial-frequency content that goes beyond the cut-o frequency of the optics. Such techniques typically must rely on extensive a priori information. In other image acquisition systems, the limiting factor may be the density of the FPA, subsequent postprocessing requirements, or transmission bitrate constraints that require data compression. We refer to the process of overcoming the limitations of the FPA in order to obtain the full resolution aorded by the selected optics as detector SR. Note that some methods may seek to perform both optical and detector SR. Detector SR algorithms generally process a set of lowresolution aliased frames from a video sequence to produce a high-resolution frame. When subpixel relative motion is present between the objects in the scene and the detector array, a unique set of scene samples are acquired for each frame. This provides the mechanism for eectively increasing the spatial sampling rate of the imaging system without reducing the physical size of the detectors. With increasing interest in surveillance and the proliferation of digital imaging and video, SR has become a rapidly growing eld. Recent advances in SR include innovative algorithms, generalized methods, real-time implementations, and novel applications. The purpose of this special issue is to present leading research and development in the area of super-resolution for digital video. Topics of interest for this special issue include but are not limited to:

Detector and optical SR algorithms for video Real-time or near-real-time SR implementations Innovative color SR processing Novel SR applications such as improved object detection, recognition, and tracking Super-resolution from compressed video Subpixel image registration and optical ow Authors should follow the EURASIP JASP manuscript format described at http://www.hindawi.com/journals/asp/. Prospective authors should submit an electronic copy of their complete manuscript through the EURASIP JASP manuscript tracking system at http://www.hindawi.com/mts/, according to the following timetable:

Manuscript Due Acceptance Notication Final Manuscript Due Publication Date

September 1, 2006 February 1, 2007 April 15, 2007 3rd Quarter, 2007

GUEST EDITORS: Russell C. Hardie, Department of Electrical and Computer Engineering, University of Dayton, 300 College Park, Dayton, OH 45469-0026, USA; rhardie@udayton.edu Richard R. Schultz, Department of Electrical Engineering, University of North Dakota, Upson II Room 160, P.O. Box 7165, Grand Forks, ND 58202-7165, USA; RichardSchultz@mail.und.nodak.edu Kenneth E. Barner, Department of Electrical and Computer Engineering, University of Delaware, 140 Evans Hall, Newark, DE 19716-3130, USA; barner@ee.udel.edu

Hindawi Publishing Corporation http://www.hindawi.com

EURASIP JOURNAL ON APPLIED SIGNAL PROCESSING

Special Issue on Advanced Signal Processing and Computational Intelligence Techniques for Power Line Communications
Call for Papers
In recent years, increased demand for fast Internet access and new multimedia services, the development of new and feasible signal processing techniques associated with faster and low-cost digital signal processors, as well as the deregulation of the telecommunications market have placed major emphasis on the value of investigating hostile media, such as powerline (PL) channels for high-rate data transmissions. Nowadays, some companies are oering powerline communications (PLC) modems with mean and peak bit-rates around 100 Mbps and 200 Mbps, respectively. However, advanced broadband powerline communications (BPLC) modems will surpass this performance. For accomplishing it, some special schemes or solutions for coping with the following issues should be addressed: (i) considerable dierences between powerline network topologies; (ii) hostile properties of PL channels, such as attenuation proportional to high frequencies and long distances, high-power impulse noise occurrences, time-varying behavior, and strong inter-symbol interference (ISI) eects; (iv) electromagnetic compatibility with other well-established communication systems working in the same spectrum, (v) climatic conditions in dierent parts of the world; (vii) reliability and QoS guarantee for video and voice transmissions; and (vi) dierent demands and needs from developed, developing, and poor countries. These issues can lead to exciting research frontiers with very promising results if signal processing, digital communication, and computational intelligence techniques are effectively and eciently combined. The goal of this special issue is to introduce signal processing, digital communication, and computational intelligence tools either individually or in combined form for advancing reliable and powerful future generations of powerline communication solutions that can be suited with for applications in developed, developing, and poor countries. Topics of interest include (but are not limited to)
Multicarrier, spread spectrum, and single carrier tech

Channel coding and equalization techniques Multiuser detection and multiple access techniques Synchronization techniques Impulse noise cancellation techniques FPGA, ASIC, and DSP implementation issues of PLC modems Error resilience, error concealment, and Joint sourcechannel design methods for video transmission through PL channels Authors should follow the EURASIP JASP manuscript format described at the journal site http://asp.hindawi.com/. Prospective authors should submit an electronic copy of their complete manuscripts through the EURASIP JASP manuscript tracking system at http://www.hindawi.com/mts/, according to the following timetable:

Manuscript Due Acceptance Notication Final Manuscript Due Publication Date

October 1, 2006 January 1, 2007 April 1, 2007 3rd Quarter, 2007

GUEST EDITORS: Moiss Vidal Ribeiro, Federal University of Juiz de Fora, Brazil; mribeiro@ieee.org Lutz Lampe, University of British Columbia, Canada; lampe@ece.ubc.ca Sanjit K. Mitra, University of California, Santa Barbara, USA; mitra@ece.ucsb.edu Klaus Dostert, University of Karlsruhe, Germany; klaus.dostert@etec.uni-karlsruhe.de Halid Hrasnica, Dresden University of Technology, Germany hrasnica@ifn.et.tu-dresden.de

niques Channel modeling

Hindawi Publishing Corporation http://asp.hindawi.com

EURASIP JOURNAL ON APPLIED SIGNAL PROCESSING

Special Issue on Numerical Linear Algebra in Signal Processing Applications


Call for Papers
The cross-fertilization between numerical linear algebra and digital signal processing has been very fruitful in the last decades. The interaction between them has been growing, leading to many new algorithms. Numerical linear algebra tools, such as eigenvalue and singular value decomposition and their higher-extension, least squares, total least squares, recursive least squares, regularization, orthogonality, and projections, are the kernels of powerful and numerically robust algorithms. The goal of this special issue is to present new ecient and reliable numerical linear algebra tools for signal processing applications. Areas and topics of interest for this special issue include (but are not limited to):
Singular value and eigenvalue decompositions, in-

Authors should follow the EURASIP JASP manuscript format described at http://www.hindawi.com/journals/asp/. Prospective authors should submit an electronic copy of their complete manuscript through the EURASIP JASP manuscript tracking system at http://www.hindawi.com/mts/, according to the following timetable:

Manuscript Due Acceptance Notication Final Manuscript Due Publication Date

October 1, 2006 February 1, 2007 May 1, 2007 3rd Quarter, 2007

cluding applications.
Fourier, Toeplitz, Cauchy, Vandermonde and semi

separable matrices, including special algorithms and architectures. Recursive least squares in digital signal processing. Updating and downdating techniques in linear algebra and signal processing. Stability and sensitivity analysis of special recursive least-squares problems. Numerical linear algebra in: Biomedical signal processing applications. Adaptive lters. Remote sensing. Acoustic echo cancellation. Blind signal separation and multiuser detection. Multidimensional harmonic retrieval and direction-of-arrival estimation. Applications in wireless communications. Applications in pattern analysis and statistical modeling. Sensor array processing.

GUEST EDITORS: Shivkumar Chandrasekaran, Department of Electrical and Computer Engineering, University of California, Santa Barbara, USA; shiv@ece.ucsb.edu Gene H. Golub, Department of Computer Science, Stanford University, USA; golub@sccm.stanford.edu Nicola Mastronardi, Istituto per le Applicazioni del Calcolo Mauro Picone, Consiglio Nazionale delle Ricerche, Bari, Italy; n.mastronardi@ba.iac.cnr.it Marc Moonen, Department of Electrical Engineering, Katholieke Universiteit Leuven, Belgium; marc.moonen@esat.kuleuven.be Paul Van Dooren, Department of Mathematical Engineering, Catholic University of Louvain, Belgium; vdooren@csam.ucl.ac.be Sabine Van Huffel, Department of Electrical Engineering, Katholieke Universiteit Leuven, Belgium; sabine.vanhuel@esat.kuleuven.be

Hindawi Publishing Corporation http://www.hindawi.com