11 views

Uploaded by Nguyễn Ngọc Tú

- color steganography
- Most Suited Mother Wavelet for Localization of Transmission Line Faults
- A Novel Method For Image Enhancement By Using SWT And DWT
- applsci-09-00259.pdf
- speech compression
- An Introduction to Discrete Wavelet Transforms
- EEE PROJECT
- 503-192
- IRJET-To Study the Mechanical Properties of Aluminum Alloy AA-6061 Welded using TIG Welding Process at Different Welding Current
- 38 (1)
- 05968955
- Adaptive Image Resolution Enhancement by using Multi Resolution Transform
- IETEJRes55110-1602238_042702.pdf
- Finger Print Recognition Using Discrete Wavelet Transform_articulo_5p
- pxc3876666
- Artifact Cancellation From Electrocardiogram by Mixed Wavelet-ICA Filter_2006
- Ch 4201557563
- [Introduction] An Introduction to Wavelets
- Image Denoising of various images Using Wavelet Transform and Thresholding Techniques
- 18 Syed Kamruddin Ahamed Final Paper

You are on page 1of 27

Yue Lu and Minh N. Do IP EDICS: 2-WAVP (Wavelets and Multiresolution Processing).

Abstract Directional information is an important and unique feature of multidimensional signals. As a result of a separable extension from one-dimensional (1-D) bases, multidimensional wavelet transforms have very limited directionality. Furthermore, different directions are mixed in certain wavelet subbands. In this paper, we propose a simple directional extension for wavelets (DEW) that xes this subband mixing problem and improves the directionality. The building block of the DEW is a two-channel 2-D lter bank with a checkerboard-shaped frequency partition. The DEW works with both the critically-sampled wavelet transform as well as the undecimated wavelet transform. In the 2-D case, it further divides the three wavelet subbands (i.e. horizontal, vertical, and diagonal) at each scale into six ner directional subbands. The DEW itself is critically-sampled, and hence will not increase the redundancy of the overall transform. Though nonseparable in essence, the proposed DEW has an efcient implementation that only requires 1-D ltering. Meanwhile, the DEW can be easily generalized to higher dimensions. In a nutshell, the proposed directional extension provides an optional tool to efciently enhance the directionality of multidimensional wavelet transforms. Numerical experiments show that certain wavelet-based image processing applications will benet from this improved directionality. Index Terms Wavelet transform, directional information, checkerboard lter bank, lter design, multidimensional signal processing, image denoising, feature extraction.

Y. Lu is with the Department of Electrical and Computer Engineering and the Coordinated Science Laboratory, University of Illinois at Urbana-Champaign, Urbana IL 61801 (e-mail: yuelu@uiuc.edu; WWW: http://www.ifp.uiuc.edu/yuelu). M. N. Do is with the Department of Electrical and Computer Engineering, the Coordinated Science Laboratory, and the Beckman Institute, University of Illinois at Urbana-Champaign, Urbana IL 61801 (e-mail: minhdo@uiuc.edu; WWW: http://www.ifp.uiuc.edu/minhdo). This work was supported by the National Science Foundation under Grant CCR-0237633 (CAREER).

April 1, 2005 DRAFT

I. I NTRODUCTION Directional information is a unique feature of multidimensional (MD) signals. Recently, the importance of directional information has been recognized by many image processing applications, including feature extraction, enhancement, denoising, classication, and compression. The wavelet transform [1], [2], [3] has a long and successful history as an efcient image processing tool. However, as a result of a separable extension from one-dimensional (1-D) bases, wavelets in higher dimensions can only capture very limited directional information. For instance, 2-D wavelets only provide three directional components, namely horizontal, vertical, and diagonal. Furthermore, the 45 and 135 directions are mixed in diagonal subbands. There have been a number of approaches in providing ner directional decomposition. Some notable examples include 2-D Gabor wavelets [4], the steerable pyramid [5], the directional lter bank [6], 2-D directional wavelets [7], brushlets [8], complex wavelets [9], [10], [11], curvelets [12], and contourlets [13], [14]. However, the wavelet transform is still very attractive for image processing applications for a number of reasons. First, the wavelet transform has a critically-sampled implementation, and can be easily extended to multidimensional cases. In contrast, most of the approaches mentioned above are expansive systems with various redundancy ratios. For example, the complex wavelet transform is 4times redundant for images, and in general 2N -times redundant for N -dimensional signals. In terms of implementation complexity, the multidimensional wavelet transforms can be implemented efciently in a separable fashion. In contrast, systems such as the directional lter bank involve nonseparable ltering and sampling and have high computational complexity. Last but not least, the theory and applications of wavelets have already been extensively studied, offering us a plethora of ready-to-use lters and processing algorithms. Therefore the natural question is: Can we extend the wavelet transform with ner directionality, while still retain its structure and desirable features? We give an afrmative answer in this paper by proposing a simple directional extension for wavelets (DEW). Based on a nonredundant checkerboard lter bank, the proposed DEW works with both the critically-sampled wavelet transform as well as the undecimated wavelet transform. In the 2-D case, the DEW leads to one lowpass subband and six directional highpass subbands at each scale, and xes the subband mixing problem of wavelets (see Figure 1). Being criticallysampled itself, the DEW will not increase the redundancy of the overall transform. Though nonseparable in essence, the DEW has an efcient implementation based on 1-D operations only. Finally, the DEW can be easily generalized to higher dimensions.

DRAFT April 1, 2005

1111111 0000000 11111111111111111111111111 00000000000000000000000000 11111111111111111111000000 1111111 0000000 111111 LH HH 00000000000000000000 HH 1111111 0000000 11111111111111111111111111 00000000000000000000000000 1111111 0000000 11111111111111111111111111 00000000000000000000000000 1111111 0000000 11111111111111111111111111 00000000000000000000000000 111111 000000 1111111 0000000 1111111 0000000 11111111111111111111000000 00000000000000000000000000 111111 000000 1111111 0000000 111111 1111111 0000000 111111

1111111 0000000 1111111 0000000 1111111 0000000 1111111 0000000 1111111 0000000 1111111 0000000 1111111 0000000 HL 1111111 0000000 1111111 0000000 1111111 0000000 1111111 0000000 1111111 0000000 1111111 0000000 1111111 0000000 1111111 0000000 1111111 0000000 1111111 0000000

(, )

LL

1111111 0000000 11111111111111111111111111 00000000000000000000000000 1111111 0000000 11111111111111111111111111 00000000000000000000000000 LH HH 00000000000000000000 HH 1111111 0000000 111111 11111111111111111111000000 11111111111111111111111111 00000000000000000000000000 1111111 0000000 0000000 1111111 11111111111111111111111111 00000000000000000000000000 00000000000000000000111111 1111111 (,0000000 ) 11111111111111111111000000

111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 HL 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000 111111 000000

4 3 1111 0000 1111 0000 1 1111 0000 1111 0000 1111 0 0000 1111 0000 4 3 1111 0000 1111 0000 1111 0000 1111 0000 111111 000000 0000 1111 0000 1111 0000 1111 1111 111111 0000 2 000000 111111 000000 1 6 111111 000000 111111 000000 1111 111111 0000 000000 1111 0000 5 111111 000000 111111 1111 000000 0000 111111 1111 0000 111111 000000 111111 000000 000000 111111 1111 000000 1111 0000 (, ) 0000

111111 000000 1111 111111 0000 1 000000 1111 111111 000000 2 0000 6 5 111111 000000 111111 000000 1111 111111 0000 000000 1111 0000 111111 000000 111111 1111 000000 0000 111111 1111 0000 111111 000000 111111 000000 000000 1111 111111 0000 000000 1111 1111 0000 0000 1111 0000

(, )

(a)

(b)

Fig. 1. Division of the 2-D frequency spectrum. (a) The frequency decomposition of the wavelet transform. (b) The frequency decomposition of the wavelet transform with the proposed directional extension.

The outline of the paper is as follows. Section II presents the lter bank construction of the directional extension in 2-D, with emphasis on intuition and illustration. The generalization of the results to higher dimensional cases is given in Section III with a more rigorous treatment. Section IV discusses lter design and efcient implementation. We will present some numerical results in Section V and conclude the paper in Section VI. Some preliminary results of this paper were reported in an earlier conference paper [15].

IN THE

2-D C ASE

The traditional way to construct 2-D wavelets is to use tensor products of their 1-D counterparts. The advantage of this approach is its simple separable implementation. Unfortunately, this also imposes serious limits on the directionality of the resulting frequency partitioning. As shown in Figure 1(a), the 2-D wavelet transform produces one lowpass subband (LL), and three highpass subbands (HL, LH, HH), corresponding to the horizontal, vertical, and diagonal directions. Furthermore, diagonal subbands mixes the directional information oriented at 45 and 135 . The main idea here is to nd some directional extension to further divide each highpass subband of the wavelets into two branches. In particular, we want to have a system with the frequency partitioning shown in Figure 1(b), which contains six directional subbands roughly oriented at 15 , 45 , 75 , 105 , 135 and 165 . This is the same frequency decomposition provided by the 2-D dual-tree complex wavelet transform [9], [10], [11], which has been shown to be successful in several image processing applications. However, the 2-D complex wavelet transform is 4-times redundant and uses a different lter bank structure compared to wavelets. In the following, we will discuss our directional extension for wavelets in two cases, i.e., when the wavelet transform is critically-sampled or undecimated.

April 1, 2005 DRAFT

4

F0 (z2 ) F0 (z1 )

D2 D1

F1 (z2 )

LL LH HL HH

D2

F0 (z2 ) F1 (z1 )

D2 D1

F1 (z2 )

horizontal vertical

D2

Fig. 2. The lter bank construction of the 2-D DWT for one level of decomposition. The ltering and downsampling operations are performed in a separable fashion. F0 and F1 denote the 1-D lowpass and highpass lters, with dark regions representing the ideal passband. D 1 = diag(2, 1) and D 2 = diag(1, 2) are the 2-fold downsampling matrices along the horizontal and vertical directions, respectively.

A. Working with the Critically-Sampled Discrete Wavelet Transform (DWT) This case corresponds to the most widely-used wavelet implementation in which the lowpass and highpass lters are always followed by a 2-fold downsampling. Figure 2 shows the lter bank implementation of the DWT for one level of decomposition. For notations used in the gure and hereafter, we use lower-case letters x[n], where n = (n1 , n2 )T , to denote 2-D discrete signals or lters. We use the corresponding uppercase letters X(z1 , z2 ) and X(ej1 , ej2 ) for their z -transforms and discrete-time Fourier transforms, respectively. When the signal or lter is 1-D, the above notations will be simplied as

x[ni ], X(zi ) and X(eji ), with i = 1 or 2 specifying the particular dimension. We use the downsampling

matrices

D 1 = diag(2, 1) = 2 0 0 1

1 0 0 2

and D 2 = diag(1, 2) =

to represent the 2-fold downsampling operations along the horizontal and vertical directions, respectively. In general, the multidimensional (MD) downsampling operation [16] is specied by an integer matrix

M as y[n] = x[M n], where x[n] and y[n] are the MD input and output signals.

To see what the desired directional extension should be, we can examine the frequency contents in the wavelet subbands. For simplicity, we rst consider the case of using ideal lters, where the frequency response of each lter is constant in the passband and exactly zero in the stopband. We know the diagonal subband (HH) captures certain directional highpass frequency information (illustrated as regions

{a, b, c, d} in Figure 3(a)) in the input image, where {a, d} and {b, c} correspond to directional information

oriented at 45 and 135 , respectively. With the downsampling operations in the wavelet transform, these

DRAFT April 1, 2005

1111111 0000000 1111111 0000000 a 1111111 0000000 1111111 0000000 1111111 0000000 1111111 0000000

111111 000000 111111 000000 111111 000000 b 111111 000000 111111 000000 111111 000000

(, ) d 1

2 c

(, )

1 b (, ) (b) a

1111111 0000000 1111111 0000000 c 1111111 0000000 1111111 0000000 0000000 1111111 1111111 (,0000000 )

111111 000000 111111 000000 111111 000000 d 111111 000000 000000 111111 000000 111111

(a) Fig. 3.

(a) The diagonal highpass frequency regions of the input signal. (b) The frequency contents of the diagonal subband

H0 (z1 , z1 ) G0 (z1 , z2 )

D1

H1 (z1 , z2 )

D1

G1 (z1 , z2 )

D1

D1

Fig. 4.

The two-channel 2-D lter bank with a checkerboard-shaped frequency partition. The dark regions represent the ideal

passband.

frequency regions will be scrambled and mapped to the actual frequency contents in the HH subband, as shown in Figure 3(b). Now to separate regions {a, d} from {c, b}, we can see that a natural choice is to use a two-channel 2-D lter bank with a checkerboard-shaped frequency partition, illustrated in Figure 4. The downsampling matrix used in the lter bank is the simple diagonal matrix D 1 .

1

Similarly, it can be checked that the same checkerboard lter bank can also be used to divide the other two wavelets subbands (HL and LH). This suggests that we can use the checkerboard lter bank as the building block for a directional extension for the DWT to improve its directionality, as shown in Figure 5, where only the analysis part is given. The original wavelet transform is kept in the rst two levels. In the third level, the three highpass subbands are further split by the checkerboard lter bank. Applying the multirate identities [16] in multirate signal processing, we can get the equivalent lter of each subband and verify that the system indeed achieves the desired frequency partitioning (Figure 1(b)) with ideal lters. Since each individual component of the proposed system, i.e., the DWT and the checkerboard lter

DRAFT

April 1, 2005

0 LL

D1

1 2 3 4 5 6

LH

D1 D1 D1 D1 D1

WT

HL HH

directional extension

Fig. 5. The lter bank construction of the proposed system for one level of decomposition. The system can be iterated on the lowpass subband 0 for multiple levels of decomposition. The synthesis part, i.e., the inverse transform, is given by a concatenation of the synthesis part of the checkerboard lter banks and the inverse wavelet transform.

bank, is critically sampled, the overall system is also critically sampled. Furthermore, if we design the checkerboard lter bank to be perfect reconstruction, then the whole system is also perfect reconstruction. A nice property of the wavelet transform is that it has an efcient separable implementation. As shown in Section IV, the proposed directional extension also has an efcient 1-D implementation in the polyphase domain, and hence will only moderately increase the computational complexity of the wavelet transform. We recognize that the proposed system has a fundamental limit in its frequency response, when we use non-ideal lters. Consider the chain of operations in a certain branch, say subband 5, in Figure 5.

F1 (z1 ) D1 F1 (z2 ) D2 H0 (z1 , z2 ) D1

Here, F1 (z1 ), F1 (z2 ) are the wavelet highpass lters along the horizontal and vertical directions, and

H0 (z1 , z2 ) is one of the checkerboard lters. Using the multirate identities, we can rewrite this branch

DRAFT April 1, 2005

1 0.8 0.6 0.4 0.2 0 2 1 0 1 0.8 0.6 0.4 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 1 1

1 0.5 0 1 0.8 0.6 0.4 0 0.2

1 0.8 0.6 0.4 0.2 0 0.2 2 1 0 1 0.8 0.6 0.4 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 1 1

0.2

0.2

0.2

The magnitude frequency responses of F1 (ej1 )F1 (ej2 ), H0 (ej21 , ej22 ) and Feq (ej1 , ej2 ) for the DWT case.

Feq (ej1 , ej2 ) = F1 (ej1 )F1 (ej2 )H0 (ej21 , ej22 ).

(1)

In Figure 6(a) and Figure 6(b), we show the typical magnitude frequency responses of some non-ideal

F1 (ej1 )F1 (ej2 ) and H0 (ej21 , ej22 ). The magnitude frequency response of the equivalent lter Feq is

shown in Figure 6(c) as the multiplication of the two previous responses. We can see that |Feq (ej1 , ej2 )| is concentrated mostly along the 135 direction. However, there are also some unwanted bumps (aliasing components) along the 45 direction. A similar phenomenon was mentioned in [11] for a nonredundant construction of the complex wavelet transform. Similar to the observations made there, it is only possible to reduce the area of those bumps by using longer lters with sharper cut-off behavior; however, the height of the bumps cannot be reduced. It is expected that this aliasing problem will have a negative impact on the performance of the proposed system. However, as shown in Section V, the overall system still benets from the improved directionality in some applications, since the equivalent frequency responses are still mostly concentrated along single directions. B. Working with the Undecimated Wavelet Transform (UWT) Another useful wavelet transform is called the undecimated wavelet transform [17], [2], which discards ` all the downsampling operations in the DWT and use appropriately upsampled lters via the algorithme a trous [17]. With J denoting the number of decomposition levels, the UWT is a (3J +1)-times redundant transform in 2-D. Note that the lowpass/highpass lters in coarser levels are upsampled accordingly so that the overall equivalent lters are the same as those from the DWT. Similar to the discussion above, we can improve the directionality of the UWT by attaching checkerboard lter banks (directional extension) to its highpass branches. The construction is similar to the one shown in Figure 5, with the DWT replaced

April 1, 2005 DRAFT

1 0.8 0.6 0.4 0.2 0 0.2 2 1 0 1 0.8 0.6 0.4 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 1 1

0.2

0.2

The magnitude frequency responses of H0 (ej1 , ej2 ) and Feq (ej1 , ej2 ) for the UWT case.

by the UWT in the wavelet part. Since the checkerboard lter banks are still downsampled, the directional extension will not increase the redundancy of the overall system.2 For the UWT case, the frequency aliasing problem is eliminated. For example, the chain of operations in subband 5 of Figure 5 will now be

F1 (z1 ) F1 (z2 ) H0 (z1 , z2 ) D1

Feq (ej1 , ej2 ) = F1 (ej1 )F1 (ej2 )H0 (ej1 , ej2 ).

Compared with (1), here we are using H0 (ej1 , ej2 ) (Figure 7(a)) instead of its upsampled version

H0 (ej21 , ej22 ) (Figure 6(b)) to divide the wavelet subband. This makes a big difference when we use lters with non-ideal frequency responses. As shown in Figure 7(b), the frequency response Feq (ej1 , ej2 )

for the UWT case is single directional and free of the unwanted aliasing components.

C. Connections with Other Related Systems As mentioned before, the 2-D wavelet transform with the proposed DEW achieves the same frequency partitioning as that from the 2-D dual-tree complex wavelet transform (CWT). However, there are several differences that are important to mention. The CWT produces complex coefcients and hence provides the phase information, which has be shown to be very useful for some signal processing applications.

2

April 1, 2005

DRAFT

As a price to pay for that, the CWT is an expansive system, being 4-times redundant for images, and in general 2N -times redundant for N -dimensional signals. In contrast, the proposed DEW is a real-valued transform, and hence discards the phase information. However, it can be nonredundant (when working with the DWT) for arbitrary dimensions. More importantly, the scaling lters in the two trees of the CWT have to satisfy the half-sample delay condition, and hence cannot be arbitrarily chosen. This leads to new challenging joint lter design problems. However, our proposed DEW can work with any existing wavelet lters to suit different applications. No lter design on the wavelet part is needed. Actually this difference reects the philosophy of this work: we try to improve the directionality of wavelets in an efcient and unintrusive way. Any existing software or hardware implementation of the wavelet transform can continue to be used. Just attach the directional extension! In [18], [19], Fernandes et al. proposed a new non-redundant version of the complex wavelet transform through either a pre-mapping or a post-mapping step. In that interesting construction, the phase information can be obtained without sacricing in redundancy. One major difference between the two approaches is that the inverse pre-projection and post-projection steps need to use innite impulse response (IIR) lters to achieve perfect reconstruction. In contrast, our proposed system has perfect reconstruction with linear-phase nite impulse response (FIR) lters.

III. G ENERALIZATION

TO

H IGHER D IMENSIONS

In this section, we generalize the proposed directional extension for wavelets to arbitrary dimensions. Let us rst take a look at the 3-D case. We know the 3-D discrete wavelet transform gives one lowpass subband and seven highpass subbands. Just like its 2-D counterpart, it suffers from limited directionality and the subband mixing problems. Figure 8 shows that the situation here is actually even more severe. Analog to the 2-D case, our idea is to nd a DEW to improve the directional selectivity of these highpass subbands.

A. Partitioning the N -Dimensional Frequency Spectrum We start with some notations. Let B denote the set of frequencies in [, ). The frequency spectrum

+ of N -dimensional (N -D) discrete signals is the Cartesian product B N = B B . . . B . Let B0 = + [0, /2), B0 = [/2, 0), B1 = [/2, ), and B1 = [, /2). One can easily verify that the N

one-level separable N -D wavelet transform (both the DWT and the UWT) decomposes B N into 2N

April 1, 2005 DRAFT

10

(a)

(b)

Fig. 8. Frequency-domain support of two 3D-DWT subbands. Dark regions represent the ideal passband. In the extreme case, information corresponding to four completely different directions is mixed in subband (a).

subbands

BN =

ki {0,1}

def

def

ki {0,1}

(2)

3

+ where Bki = Bki Bki for i = 1, . . . , N . In other words, if we calculate the equivalent lter

of each

wavelet subband, the frequency-domain support of that lter will be one of the 2N regions in (2). For

each highpass subband Bk1 Bk2 . . . BkN with at least one ki = 1, we can expand it using the distributive

law

+ + + Bk1 Bk2 . . . BkN = (Bk1 Bk1 )(Bk2 Bk2 ) . . . (BkN BkN )

def

=

si {+,}

(3) (4)

=

si {+,}

We group the 2N terms in (3) into 2N 1 different polarized pairs in (4), so that each of them corresponds to the frequency support of some real-valued lter. For the 3-D case, we show in Figure 9(a) and Figure 9(b) two possible subbands from (4). We can see that each of them corresponds to a single direction, which is in sharp contrast to the situation in Figure 8(a). In addition to the full splitting in (4), it is also possible to have other partial splitting. For example,

Bk1 Bk2 . . . BkN = s1 s n+1 n+1 Bk1 . . . Bkn+1 Bk1 1 . . . Bkn+1 Bkn+2 . . . BkN , s s

(5)

si {+,}

where n = 1, . . . , N 1. The full splitting is a special case when n = N 1. Figure 9(c) shows one of the possible subbands from the partial splitting in 3-D. These subbands are useful, since they

3

April 1, 2005

DRAFT

11

(a) Fig. 9.

(b)

(c)

Frequency-domain support of several subbands, with dark regions representing the ideal passbands. Part (a) and Part

(b): Two possible subbands from the full splitting in (4). Each of them corresponds to a single direction, which is in sharp contrast to the situation in Figure 8(a). Part (c): One possible subband from the partial splitting in (5).

correspond to lters that can capture signals with singularities lying on some low-dimensional manifold structures (e.g. curves, surfaces, etc.) [20]. B. The Directional Extension for Wavelets in N -D The main result in this section is that we can achieve the subband splitting in (4) and (5) by attaching some directional extension to the highpass subbands of the wavelet transform. The basic building blocks of the N -D DEW are checkerboard lter banks shown in Figure 10(a). For notations, we use CBi,j to denote a checkerboard lter bank whose ltering operations are carried out along the ith and j th dimensions. The downsampling matrix is dened to be D i = diag(1, . . . , 1, 2, 1, . . . , 1), which keeps every other sample along the ith dimension. The two channels of the (analysis) lter bank are denoted by CBi,j (0) and CBi,j (1), respectively. Lemma 1: Suppose we attach the lter bank CBi,i+1 (1 i N 1) to one of the highpass subbands of the N -D wavelet transform (either the DWT or the UWT) with an ideal frequency sup port Bk1 Bk2 . . . BkN . If all lters used are ideal, the wavelet subband will be split into two channels + + + with passband support Bk1 . . . Bki1 (Bki Bki+1 Bki Bki+1 )Bki+2 . . . BkN and Bk1 . . . Bki1 (Bki Bki+1 + Bki Bki+1 )Bki+2 . . . BkN , respectively. i1

Remark 1: This is the generalization of the 2-D result in Section II and is proved by applying the multirate identities. See Appendix for details. As shown in Figure 10(b), an n-level DEW is a binary tree-structured expansion of lter banks

CB1,2 , CB2,3 , . . ., and CBn,n+1 , for which we have the following theorem:

Theorem 1: Suppose we attach the n-level DEW (1 n N 1) to one of the highpass subbands of

the N -D wavelet transform (either the DWT or the UWT) with an ideal frequency support Bk1 Bk2 . . . BkN . April 1, 2005 DRAFT

12

H0 (zi , zj ) Di CBn,n+1

0

0

CB2,3 CB1,2

0 1 0 1

CBn,n+1

H1 (zi , zj ) Di CBi,j

CB2,3

2n subbands

0 1

Fig. 10. (a) CBi,j is the analysis part of the checkerboard lter bank. The subscripts i, j (i = j) specify that the lters H0 (zi , zj ) and H1 (zi , zj ) operate along the ith and jth dimensions. (b) The n-level DEW is a binary tree-structured concatenation of checkerboard lter banks CB1,2 , CB2,3 , . . ., and CBn,n+1 . Notice the arrangement of subscripts for the simple expansion rule.

If all lters used are ideal, the wavelet subband will be split into 2n channels as in (5). Remark 2: This is proved by successively applying the result in Lemma 1. See Appendix for details. Note that the partial splitting in (5) is along dimensions 1, 2, . . . , n+1. However, through a permutation of indices, we can construct a similar n-level DEW to generate partial splitting along n + 1 arbitrary dimensions d1 , d2 , . . . , dn+1 , where di {1, . . . , N }. Moreover, the full splitting in (4) is a special case when n = N 1. In some applications, e.g., enhancing a particular direction in the signal, one might be only interested in getting a single frequency region. Instead of using the full binary tree expansion of 2n 1 checkerboard lter banks in Figure 10(b) to get all 2n subbands, we can use a connected sequence of n checkerboard lter banks for that specic subband. Such a sequence of checkerboard lter banks is specied by the following proposition, which can be veried in the proof of Theorem 1 (Appendix B).

s1 s n+1 n+1 Proposition 1: Suppose the subband of interest is Bk1 . . . Bkn+1 Bk1 1 . . . Bkn+1 Bkn+2 . . . BkN , s s

where si {+, }. The desired adaptive DEW for that subband will be

CB1,2 (c1 ) CB2,3 (c2 ) . . . CBn,n+1 (cn ),

where CBi,i+1 is connected to channel # ci1 (either 0 or 1) of CBi1,i (i = 2, . . . , n) with the following expansion rule: for DWT:

ci = 1 (ki ki+1 ), (ki ki+1 ),

; ci =

0, 1,

if si = si+1 if si = si+1

DRAFT April 1, 2005

13

H0 (z) x H1 (z)

D1

y0

D1

G0 (z) x

D1

y1

D1

G1 (z)

Fig. 11.

A maximally decimated two-channel 2-D lter bank with analysis lters H0 (z) (z = (z1 , z2 )), H1 (z), synthesis

def

lters G0 (z), G1 (z), and the downsampling matrix D 1 . x and x are the input and reconstructed signals, respectively. y0 and y1 are the two output signals of the analysis part. y0 H (z) D1 y1 G(z) D1

x

1 z1

D1

D1 z1

Fig. 12.

The equivalent polyphase form of the lter bank in Figure 11. H (z) and G(z) are 2-by-2 polyphase matrices.

IV. F ILTER

As stated before, there is no need to modify the wavelet transform part and we can choose any existing wavelet lters to suit different applications. The remaining task is to design a 2-D checkerboard lter bank with the desired frequency response and perfect reconstruction. Here we will present a design based on a parameterization of the polyphase matrices. The advantage of this design is that the resulting lter banks have efcient implementation that only requires 1-D ltering. A more general design of checkerboard lter banks, in particular the ones that are tailored to work in combination with wavelets, will be investigated in a future work. A. Design by Polyphase Matrices Parameterization Let us consider a maximally decimated two-channel lter bank in 2-D with the downsampling matrix

D 1 , shown in Figure 11. The well-known equivalent polyphase form of the system is given in Figure 12.

To simplify notation, we denote z = (z1 , z2 ). The relationship between the analysis and synthesis lters

{Hk (z), Gk (z)} and the polyphase matrices H(z) and G(z) can be expressed as [16]

1 2 2 Hk (z) = H k,0 (z1 , z2 ) + z1 H k,1 (z1 , z2 ) 2 2 Gk (z) = G0,k (z1 , z2 ) + z1 G1,k (z1 , z2 ),

(6)

April 1, 2005

DRAFT

14

for k = 0, 1. We can see from Figure 12 that a sufcient condition for the lter bank to have perfect reconstruction is

H(z) G(z) = I,

(7)

where I is the identity matrix. In our design, we choose H(z) and G(z) to be 0.5 0 1 (z) H(z) = 2 0.5(z) 1 0 z1 and

1 G(z) = 2 0

1 1 z1 (z) 1 z1

(8)

2 0 (z) 1

where (z) is a free design parameter. It is easy to verify that the perfect reconstruction condition (7) is structurally guaranteed for arbitrary choice of (z). This form of polyphase parameterization was rst proposed by Phoong et al. [21] for the 1-D lter bank. However, the 1-D to 2-D mapping method proposed in that work can only be used to design 2-D lters banks with a single parallelogram-shaped support, such as the diamond shape. While the checkerboard shape we want here does not belong to this class. In the following, we will propose a novel 1-D to 2-D mapping for the checkerboard lter bank in the polyphase domain. Substituting (8) into (6), we get

H0 (z) =

1 2 1 + z1 (z1 , z2 ) 2

(9)

Similarly, we can also write down H1 (z), G0 (z) and G1 (z). Actually, they are all related to H0 (z) as follows.

H1 (z) = z1 2 2H0 (z) 1 H0 (z)

If H0 (z) is the ideal lter with the desired checkerboard-shaped frequency support shown in Figure 4, i.e., if its Fourier transform takes the constant value 2 in the passband and 0 in the stopband, we can then verify from (10) - (12) that the other three lters H1 (z), G0 (z) and G1 (z) will also achieve the desired frequency response. For instance, G1 (ej1 , ej2 ) = ej1 H0 (ej(1 ) , ej2 ), i.e., G1 (ej1 , ej2 ) is obtained by shifting H0 (ej1 , ej2 ) horizontally by and multiplying a linear phase.

DRAFT April 1, 2005

2 2 2

15

(, )

(, )

(, )

+1 1

(, )

1 +1

+j

1

j j

j +j

(, )

j +j

+j

(, )

1 2 z1 (z1 , z2 )

M (z1 )

M (z2 )

1 2 Fig. 13. The separable decomposition of z1 (z1 , z2 ). The values of the 2-D Fourier transform of the lters are shown in the

gure.

Therefore, we only need to design H0 (z) to approximate the ideal lter on the checkerboard support.

1 2 In turn, this implies that the Fourier transform of the lter z1 (z1 , z2 ) should take constant values

(1, 1, 1, 1) in the four quadrants of the 2-D frequency plane, as illustrated in Figure 13. Since this is

a separable lter, we can decompose it as the product of two 1-D lters M (z1 ) and M (z2 ), i.e.,

1 2 z1 (z1 , z2 ) = M (z1 ) M (z2 ).

(13)

If we further constrain M (z) to have real coefcients, i.e., M (ej ) = M (ej ), then one of the only two choices is

4

M (ej ) =

j, +j,

(14)

M (z) = z 1 (z 2 ),

(15)

1 2 (z) = z2 (z1 )(z2 ).

(16)

From (14) and (15), we want (z) to be an allpass lter with half-sample delay specied as:

|(ej )| = 1, (ej ) = 0.5, (, );

(17) (18)

In summary, the proposed lter bank design process is given as follows. (1) Design a lter (z) to approximate the conditions in (17) and (18).

4

DRAFT

April 1, 2005

16

(2) Obtain the analysis and synthesis lters H0 , H1 , G0 , G1 from (16) and (9) - (12). Note that step (2) is only needed for lter bank analysis. As we will see later, knowing (z) is sufcient for actual implementation (see Figure 16). B. Designing the Allpass Filter with Half-Sample Delay The half-sample delay conditions (17) and (18) can only be realized by an IIR lter with an irrational system function. There are various ways to design FIR lters to approximate the conditions. See [22] for a review. Here we use a simple approach by choosing (z) to be a type-II linear phase lter [23] of length-2d:

(z) = b0 (1 + z) + b1 (z 1 + z 2 ) + . . . + bd1 (z (d1) + z d ).

Note that the phase condition (18) is satised exactly, and we need to design the coefcients b0 , . . . , bd1 such that the magnitude of the frequency response

d1

|(e )| =

j k=0

1 2bk cos (k + ) 2

is as close to 1 as possible. In Table I, we list the coefcients of the lters (z) of different lengths designed by the ParksMcClellan algorithm. We show in Figure 14 the magnitude responses of those (z). From (16), (9), and (10), we can verify that the resulting 2-D lters H0 and H1 are of sizes (4d 1) (4d 1) and

(8d 3) (8d 3), respectively. In Figure 15, we give the magnitude response of the 2-D lter H0 (z)

obtained from a (z) with d = 6. We can see that it is a good approximation to the ideal checkerboard lter.

DRAFT April 1, 2005

17

10

0

Magnitude

1.5 1 0.5 0 1

Magnitude (dB)

20

40

0.5

50

x

60

70

0.1

0.2

0.8

0.9

Fig. 14. The magnitude frequency responses of (z) with d = 3, 4, 5, and 6. Note that the phase responses (not shown in the gure) satisfy (18) exactly. y0

Fig. 15.

x

1 z1

D1

1 2

1 2 z2 (z1 )(z2 )

D1

x z1

1 2 z2 (z1 )(z2 )

1 2 z2 (z1 )(z2 )

1 2 z2 (z1 )(z2 )

D1

y1 2z1

1 z 1 2 1

D1

G(z)

C. Efcient Implementation A nice property of the proposed lter design is that the structure is similar to a ladder network [21]. In Figure 16, we show the polyphase implementation of the analysis and synthesis parts of the lter bank. Although the designed 2-D lters are nonseparable, their polyphase components are separable. Meanwhile, the downsampling matrix D 1 = diag(2, 1) is also separable. Therefore, the entire system can be implemented by efcient 1-D operations only. To have a rough estimate of the computational complexity of the proposed DEW, we can count the number of arithmetic operations needed for each new input sample to the DEW. In 2-D, the DEW is just a single-level checkerboard lter bank implemented in the polyphase domain as in Figure 16. For simplicity, we assume the convolution operations are implemented as straight polynomial products, while similar analysis can be done for FFT-based convolutions. Suppose the lter (z) used in Figure 16 is of

April 1, 2005 DRAFT

18

TABLE II N UMBER OF ARITHMETIC OPERATIONS / INPUT SAMPLE FOR DIFFERENT SYSTEMS . B OTH THE DWT

PERFORM A AND THE

UWT

AND

LENGTH

2d.

# of multiplications/input sample DWT DEW for DWT UWT DEW for UWT ( L1 +L2 )N 2

12N J 12N

12N J 12N

length 2d. Since it is a symmetric lter, the ltering by (z1 ) requires d multiplications/input sample

2 and 2d 1 additions/input sample. Moreover, the ltering by (z2 ) can be efciently implemented in

the polyphase domain and still requires d multiplications/input sample and 2d 1 additions/input sample. We can verify that the total number of operations needed for the analysis part of the DEW is 2d + 1 multiplications/input sample and 4d 1 additions/input sample. The number of operations needed for the synthesis part is the same, since the analysis and synthesis parts are duals of each other. For the general N -dimensional case, the DEW (for full splitting) is a (N 1)-level binary tree expansion of checkerboard lter banks. The number of operations becomes (2d + 1)(N 1) multiplications/input sample and (4d 1)(N 1) additions/input sample. For comparison purposes, we show in Table II the number of arithmetic operations/input sample for the DWT, the UWT, and the corresponding DEW in either case. We assume both wavelet transforms are implemented in the polyphase domain 5 and perform a J -level decomposition of a N -dimensional signal. The lowpass and highpass lters used in the wavelet transforms are of length L1 and L2 , respectively. We do not assume special structures, such as linear phase or orthogonality, on them. The DEW is attached to all the highpass subbands of the wavelet decomposition tree. Here we just list the numbers without derivation. A detailed analysis on the computational complexity of multirate systems can be found in [2]. We can see from Table II that the relative increase in computational complexity (by attaching the DEW) is mainly determined by the ratio d/(L1 +L2 ). In 2-D and for some typical numbers L1 = 9, L2 = 7, d = 6,

5

April 1, 2005

DRAFT

19

1 0.8 0.6 0.4 0.2 0 0.2 2 0 1 0.8 0.6 0.4 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 1 1 0 1 0.2 0.8 0.6 0.4 0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 1 1

0 0.2 0.2 0 1 0.8 0.6 0.4 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 1 1

0.2

0.2

A few basis images from different transforms. (a) - (c): Some basis images in the frequency domain. (d) - (f): Some

applying the DEW to the DWT corresponds to an increase of 61% multiplications and 123% additions. Similarly, applying the DEW to the UWT brings an increase of 81% multiplications and 164% additions.

V. N UMERICAL R ESULTS All experiments in this section use the checkerboard lters designed in Section IV with a (z) of length 12 (i.e. d = 6). These checkerboard lters are close to having the ideal frequency response and thus can provide better separation of the directional information. Meanwhile, the sharp cut-off of these lters can help alleviate the aliasing problem mentioned in Section II. For the wavelet transform, we use three levels of decomposition and employ Daubechiess symmlet of length 12 [1]. As before, we use DWT and UWT to represent the critically-sampled and the undecimated wavelet transforms, respectively. Correspondingly, we use DWT-DEW and UWT-DEW to denote the two wavelet transforms equipped with the proposed directional extension.

A. Basis Images Figure 17 shows a few basis images from different transforms in both frequency and spatial domains. The basis images of UWT are not shown here, since they are equivalent to those of DWT. We can see that the basis images of DWT-DEW and UWT-DEW have better directionality than those of DWT.

April 1, 2005 DRAFT

20

(a)

(b)

(c)

(d)

Fig. 18.

(a) A synthetic image consisting of diagonal lines. (b) The diagonal subband of the DWT. (c) Subband 5 of the

As expected, the critically-sampled DWT-DEW suffers from the frequency aliasing problem, resulting in some unwanted bumps in the opposite direction (Figure 17(b)). Although the height of these bumps cannot be reduced, it is possible to make the area of the bumps small by using lters with sharp cutoff. However, we should see that most of the subband energy of DWT-DEW concentrates on a single direction, and hence it might still be useful for some image processing applications. Figure 17(c) shows that the aliasing problem is eliminated for UWT-DEW, but at the cost of a redundant system. In the spatial domain, the basis image of DWT from the diagonal (HH) subband exhibits an apparent checkerboard artifact (Figure 17(d)). In contrast, DWT-DEW (Figure 17(e)) and UWT-DEW (Figure 17(f)) succeed in isolating different directions, and no checkerboard effect appears.

B. Directional Feature Extraction We rst apply the DWT and DWT-DEW on a synthetic image (Figure 18(a)), which consists of lines oriented at both diagonal directions. The result of the UWT-DEW is not shown here, since it is very close to that of the DWT-DEW. Figure 18(b) shows the diagonal (HH) subband of the DWT. As discussed before, 45 and 135 directions are mixed in this subband and the separable wavelets cannot discriminate

DRAFT April 1, 2005

21

(a)

(b)

(c)

(d)

Fig. 19.

(a) The cameraman image. (b) The diagonal subband of the DWT. (c) Subband 5 of the DWT-DEW. (d) Subband 6

of the DWT-DEW.

between them. In Figure 18(c) and Figure 18(d), we show the subbands 5 and 6 of the DWT-DEW. Clearly the subband mixing problem is solved, and the two subbands correctly capture the corresponding directional information. We then do the same experiment on the cameraman image. As shown in Figure 19(c) and Figure 19(d), the DWT-DEW correctly captures and separates the directional information at 45 and 135 in two subbands.

C. Denoising The improvement in directionality of the transform will lead to improvements in many applications. As an example, we compare the performance of the DWT and the UWT, both with and without the proposed DEW, in image denoising. In this experiment, we apply additive white Gaussian noise of different variances 2 to the Barbara image. A simple hard thresholding scheme [24] with threshold

T = 3 is used for all 4 transforms. In Figure 20, we show the peak signal-to-noise ratio (PSNR) of

denoised Barbara images using different transforms under various noise levels. We can see that in all cases the DEW help improve the performance of the original transforms. For a noisy image with PSNR

April 1, 2005 DRAFT

22

17

18

23

24

25

Fig. 20.

PSNR (dB) of denoised Barbara images by using different transforms under various noise levels.

= 20.19 dB, the improvement (by using the proposed DEW) is about 0.30 dB for DWT, and a more signicant 1.01 dB for UWT. Figure 21 displays a zoom-in comparison of the denoising results. We can see that edges and directional textures are better restored when we apply the proposed DEW to the DWT and UWT. Despite the relatively small improvement in PSNR, the DWT-DEW clearly outperforms the DWT (especially in regions with directional textures) in terms of visual quality. VI. C ONCLUSION In this work, we proposed a simple directional extension for wavelets that equips the multidimensional wavelet transform (both the DWT and the UWT) with ner directionality. The lter bank construction of the overall system is a concatenation of the separable wavelet transform with a sequence of checkerboard lter banks. No modication on the wavelet part is necessary and hence any existing wavelet implementation can continue to be used. In 2-D, the proposed DEW further divides the three wavelet subbands (i.e. horizontal, vertical, and diagonal) at each scale into six ner directional subbands. The DEW is nonredundant, and can be easily generalized to higher dimensions. Moreover, the DEW can be optional and adaptive to specic directions. We also proposed a design of the checkerboard lter banks based on a parameterization of the polyphase matrices and a novel 1-D to 2-D mapping. The resulting lter banks have efcient implementation that only requires 1-D ltering. Numerical experiments indicate the

DRAFT April 1, 2005

23

Comparison of denoising results when applying DWT, DWT-DEW, UWT, and UWT-DEW on the Barbara image.

A PPENDIX A. Proof of Lemma 1 We assume the wavelet transform is the DWT. The proof for the case of UWT is similar. Suppose we

attach CBi,i+1 to one of the wavelet highpass subbands with an ideal frequency support Bk1 Bk2 . . . BkN .

Fk1 (z1 ) D 1 . . . FkN (zN ) D N H0 (zi , zi+1 ) D i ,

N -D DWT April 1, 2005 CBi,i+1 DRAFT

24

i+1 (, )

(, )

Fig. 22.

where Fkj with kj {0, 1} are the ideal lowpass and highpass wavelet lters, and H0 is one of the checkerboard lters in Figure 10(a). Applying the multirate identities, we can rewrite this channel as

2 2 Fk1 (z1 ) . . . FkN (zN ) H0 (zi , zi+1 )

equivalent lter Feq (z1 , . . . , zN )

D1 . . . DN Di

equivalent downsampling matrix

supp (Feq ) = supp Fk1 (ej1 ) . . . FkN (ejN ) H0 (ej2i , ej2i+1 ) = supp Fk1 (ej1 ) . . . FkN (ejN ) supp H0 (ej2i , ej2i+1 )

(19)

The frequency support of H0 (ej2i , ej2i+1 ) is shown in Figure 22. One can easily verify that B . . . B (B + B + B B )B . . . B , if ki = ki+1 k1 ki1 ki ki+1 ki ki+1 ki+2 kN supp (Feq ) = B . . . B (B + B B B + )B . . . B , if ki = ki+1 . k1 ki1 ki ki+1 ki ki+1 ki+2 kN

(20)

In either case, the two channels of CBi,i+1 split the wavelet subband into two subbands as described in Lemma 1.

B. Proof of Theorem 1 Again, we assume the wavelet transform is the DWT. The proof for the case of UWT is similar. The

n-level DEW has 2n subband channels. The chain of operations from the input to any one of the channels

is

N -D DWT Hc1 (z1 , z2 ) D 1 . . . Hcn (zn , zn+1 ) D n ,

one channel of the n-level DEW

DRAFT

April 1, 2005

25

where ci {0, 1}. Applying the multirate identities, we can get the frequency-domain support of the equivalent lter as

n supp Feq = Bk1 Bk2 . . . BkN supp Hc1 (ej21 , ej22 ) . . . Hcn (ej2n , ej2n+1 )

Bk1 Bk2 n

. . . BkN

i=1

=

i=1

Bk1 Bk2 . . . BkN supp Hci (ej2i , ej2i+1 ) Bk1 Bk2 . . . BkN supp Hcn (ej2n , ej2n+1 )

(21)

,

n1 = supp Feq

(22)

i where Feq (i = 1, . . . , n) is the equivalent lter from an i-level DEW. Note that each term in (21) is

the frequency support of a subband produced by attaching a single-level DEW (CBi,i+1 ) to the wavelet subband (e.g. see (19)). This implies that we can successively apply the one-step splitting result in Lemma 1. More formally, we will use an induction proof. (1) The claim in Theorem 1 is true for n = 1, as ensured by Lemma 1. (2) Now by induction, we suppose that the (n 1)-level DEW (2 n N 1) can split the wavelet highpass subband into 2n1 regions as

s1 sn s s Bk1 . . . Bkn Bk1 1 . . . Bkn n Bkn+1 . . . BkN ,

(23)

si {+,}

and we want to check if the splitting result works for the n-level DEW. (3) From (22), (19), and (20), each subband in (23) will be split into two in the n-level DEW:

s1 sn s s Bk1 . . . Bkn Bk1 1 . . . Bkn n Bkn+1 . . . BkN + + (Bkn Bkn+1 Bkn Bkn+1 ) ,

and

s1 sn s s Bk1 . . . Bkn Bk1 1 . . . Bkn n Bkn+1 . . . BkN + + (Bkn Bkn+1 Bkn Bkn+1 ) ,

s1 sn sn s s sn Bk1 . . . Bkn Bkn+1 Bk1 1 . . . Bkn n Bkn+1 Bkn+2 . . . BkN ,

and

s1 sn sn s s sn Bk1 . . . Bkn Bkn+1 Bk1 1 . . . Bkn n Bkn+1 Bkn+2 . . . BkN ,

s1 s n+1 n+1 Bk1 . . . Bkn+1 Bk1 1 . . . Bkn+1 Bkn+2 . . . BkN , s s

DRAFT

26

which proves the claim in the theorem for all n {1, . . . , N 1}.

R EFERENCES

[1] I. Daubechies, Ten Lectures on Wavelets. Philadelphia, PA: SIAM, 1992, notes from the 1990 CBMS-NSF Conference

on Wavelets and Applications at Lowell, MA. [2] M. Vetterli and J. Kova evi , Wavelets and Subband Coding. Prentice-Hall, 1995. c c [3] S. Mallat, A Wavelet Tour of Signal Processing. Academic Press, 1998. [4] J. Daugman, Two-dimensional spectral analysis of cortical receptive eld prole, Vision Research, vol. 20, pp. 847856, 1980. [5] E. P. Simoncelli, W. T. Freeman, E. H. Adelson, and D. J. Heeger, Shiftable multiscale transforms, IEEE Trans. Inform. Th., Special Issue on Wavelet Transforms and Multiresolution Signal Analysis, vol. 38, no. 2, pp. 587607, March 1992. [6] R. H. Bamberger and M. J. T. Smith, A lter bank for the directional decomposition of images: theory and design, IEEE Trans. Signal Proc., vol. 40, no. 4, pp. 882893, April 1992. [7] J. P. Antoine, P. Carrette, R. Murenzi, and B. Piette, Image analysis with two-dimensional continuous wavelet transform, Signal Processing, vol. 31, pp. 241272, 1993. [8] F. G. Meyer and R. R. Coifman, Brushlets: a tool for directional image analysis and image compression, Journal of Appl. and Comput. Harmonic Analysis, vol. 5, pp. 147187, 1997. [9] N. Kingsbury, Complex wavelets for shift invariant analysis and ltering of signals, Journal of Appl. and Comput. Harmonic Analysis, vol. 10, pp. 234253, 2001. [10] I. W. Selesnick, The double-density dual-tree DWT, IEEE Trans. Signal Proc., vol. 52, no. 5, pp. 13041314, May 2004. [11] R. G. Baraniuk, N. Kingsbury, and I. W. Selesnick, The dual-tree complex wavelet transform - a coherent framework for multiscale signal and image processing, IEEE SP Mag., submitted, 2004. [12] E. J. Cand` s and D. L. Donoho, Curvelets a suprisingly effective nonadaptive representation for objects with edges, e in Curve and Surface Fitting, A. Cohen, C. Rabut, and L. L. Schumaker, Eds. 1999. [13] M. N. Do and M. Vetterli, The contourlet transform: an efcient directional multiresolution image representation, IEEE Trans. Image Proc., to appear, http://www.ifp.uiuc.edu/minhdo/publications. [14] Y. Lu and M. N. Do, CRISP-Contourlets: a critically-sampled directional multiresultion image representation, in Proc. of SPIE conference on Wavelet Applications in Signal and Image Processing X, San Diego, USA, 2003. [15] , The ner directional wavelet transform, in Proc. IEEE Int. Conf. Acoust., Speech, and Signal Proc., Philadelphia, 2005. [16] P. P. Vaidyanathan, Multirate Systems and Filter Banks. Prentice Hall, 1993. [17] M. Holschneider, R. Kronland-Martinet, J. Morlet, and P. Tchamitchian, A real-time algorithm for signal analysis with the help of the wavelet transform, in Wavelets, Time-Frequency Methods and Phase Space. pp. 289297. [18] F. Fernandes, R. van Spaendonck, and C. Burrus, A new framework for complex wavelet transforms, IEEE Trans. Signal Proc., vol. 51, no. 7, pp. 1825 1837, July 2003.

DRAFT April 1, 2005

27

[19] , Multidimensional, mapping-based complex wavelet transforms, IEEE Trans. Image Proc., vol. 14, no. 1, pp. 110 124, January 2005. [20] W. Chan, H. Choi, and R. Baraniuk, Directional hypercomplex wavelets for multidimensional signal analysis and processing, in Proc. IEEE Int. Conf. Acoust., Speech, and Signal Proc., Montreal, Canada, May 2004. [21] S.-M. Phoong, C. W. Kim, P. P. Vaidyanathan, and R. Ansari, A new class of two-channel biorthogonal lter banks and wavelet bases, IEEE Trans. Signal Proc., vol. 43, no. 3, pp. 649665, Mar. 1995. [22] T. Laakso, V. Valimaki, M. Karjalainen, and U. Laine, Splitting the unit delay - tools for fractional delay lter design, IEEE SP Mag., vol. 13, no. 1, pp. 30 60, January 1996. [23] A. V. Oppenheim, R. W. Schafer, and J. R. Buck, Discrete-Time Signal Processing, 2nd ed. Prentice-Hall, 1999. [24] D. Donoho and I. Johnstone, Ideal spatial adaptation via wavelet shrinkage, Biometrika, pp. 425455, December 1994.

April 1, 2005

DRAFT

- color steganographyUploaded byBrijesh Kumar
- Most Suited Mother Wavelet for Localization of Transmission Line FaultsUploaded byIJSTR Research Publication
- A Novel Method For Image Enhancement By Using SWT And DWTUploaded byEighthSenseGroup
- applsci-09-00259.pdfUploaded byMohamed Elsayed
- speech compressionUploaded byAmol Chaudhari
- An Introduction to Discrete Wavelet TransformsUploaded byAtacan Özkan
- EEE PROJECTUploaded bySameer Hussain
- 503-192Uploaded byanithabas
- IRJET-To Study the Mechanical Properties of Aluminum Alloy AA-6061 Welded using TIG Welding Process at Different Welding CurrentUploaded byIRJET Journal
- 38 (1)Uploaded byEmon Khan
- 05968955Uploaded byRoshan Jayswal
- Adaptive Image Resolution Enhancement by using Multi Resolution TransformUploaded byParam Addala
- IETEJRes55110-1602238_042702.pdfUploaded byAbhishek Mudgal
- Finger Print Recognition Using Discrete Wavelet Transform_articulo_5pUploaded byRommel García Hernández
- pxc3876666Uploaded byismailhussain22
- Artifact Cancellation From Electrocardiogram by Mixed Wavelet-ICA Filter_2006Uploaded byMahesh Divakar
- Ch 4201557563Uploaded byAnonymous 7VPPkWS8O
- [Introduction] An Introduction to WaveletsUploaded byfor_test
- Image Denoising of various images Using Wavelet Transform and Thresholding TechniquesUploaded byAnonymous CUPykm6DZ
- 18 Syed Kamruddin Ahamed Final PaperUploaded byiiste
- novel Fractional Wavelet Transform With Closed-Form ExpressionUploaded byAdam Bachtiar Sa'ban
- An Optimal Technique for ECG Noise Reduction in Real Time ApplicationsUploaded byHoang Manh
- ecg encodingUploaded byJobins George
- Kutil08aUploaded byNeelima Vinjmuri
- 09 Chapter 3Uploaded byabhirbhandary
- ChaptersUploaded byRichurajan
- Atmospheric Correction of Remotely Sensed Images in Spatial and Transform DomainUploaded byAI Coordinator - CSC Journals
- V2I1018Uploaded byeditor_ijarcsse
- Image FusionUploaded byJournalNX - a Multidisciplinary Peer Reviewed Journal
- Image FusionUploaded byshailygoyal

- VINASAT1Uploaded byNguyễn Ngọc Tú
- Danh sachUploaded byNguyễn Ngọc Tú
- Viterbi DecoderUploaded byNguyễn Ngọc Tú
- ckUploaded byNguyễn Ngọc Tú
- ckUploaded byNguyễn Ngọc Tú
- ckUploaded byNguyễn Ngọc Tú
- Bản kế hoạch thực hiện Project IUploaded byNguyễn Ngọc Tú
- BTHDCmoiUploaded byNguyễn Ngọc Tú
- BaiTap-Chuong123Uploaded byNguyễn Ngọc Tú
- Seminar-TopicUploaded byNguyễn Ngọc Tú
- Phan cong DA thiet ke 2 (1)Uploaded byNguyễn Ngọc Tú

- 18 REB 670 IEDUploaded byPhonepasong Phonekeo
- Aix Quick SheetUploaded bySyed1122
- Sample JD - Product HeadUploaded bysanjaykrocks
- E-commerce Limitation and StudyUploaded byAnonymous MZRzaxFgVL
- Queue Type Pcq-mangle-queue Tree v2 ReloadUploaded byRafa Sena
- 09 Introduction to GENEX Probe (Ver1.4)Uploaded byRafael Andres Rodriguez Marulanda
- Web 3.0, Sematic Web, Object Level Data RetrievalUploaded byZ. Wen
- SAP Terp10 Summary Sap Terp10.Com .ArUploaded byAmy Brady
- Relayfax EnUploaded byJohn Thomas Rogan
- Clips Case StudyUploaded byCedric Dabija
- CampBuzz XXXIIUploaded byEME Alumni Association
- HS-SCCH Channel Coding Details Based on TS 25.212 Rel 9Uploaded byAshok Hs
- rocs2Uploaded bySimonferez
- Development of a Multi-Agent System for the Support of Group Decisions Utilizing Argumentation and Multicriteria MethodsUploaded byAnonymous svLu6I
- Module+2+Full+Manual.pdfUploaded byRam Rajya
- 3. Humanities-A Study of Image Processing in Analyzing Tree Ring-P. Meenakshi SundariUploaded byImpact Journals
- 4 Ways to Schedule Tasks in Spring 3 _ @Scheduled Example _ How to Do in JAVAUploaded byVishal Nigam
- Big DataUploaded byDAN
- flange designUploaded byJilu Easow Raju
- Curso SQL ORACLE02Uploaded bycarlos
- Pass Thru Pro II _ Car ECU Programming _ Snap-On DiagnosticsUploaded byPepenauta Internauta
- fsp-3000-agileconnectUploaded byJose Valenzuela
- DSF_LAB MANUAL(SE IT)Uploaded byAvi Surywanshi
- AttrMappingUploaded byHassan Ogloo
- UnixUploaded bypooh4u_291
- uTorrent Disable AdsUploaded byFoxman2k
- pagesUploaded by1981todurkar
- CDU V2 InstructionsUploaded byNv. Luc
- OMRON MANUAL SERIES NV PTUploaded byFernando Jaime Alonso Martínez
- 00797315Uploaded byJohn McClane