Design of Navel Adaptive TDBLMS-based Wiener Parallel to TDBLMS Algorithm for Image Noise Cancellation

Design of Navel Adaptive TDBLMS-based Wiener Parallel to TDBLMS Algorithm for Image Noise Cancellation

Dinesh Yadav1, Ajay Boyat2

1,2Department of Electronics and Communication Medi-caps Institute of Technology and Management, Indore, MP, India

Abstract In this paper, we proposed a navel digital adaptive algorithm that filtered highly contaminated noisy images. The new algorithm so called as Design of A Navel Adaptive TDBLMSBased Wiener parallel to TDBLMS algorithms for Image Noise Cancellation. To improve the quality of images, adaptive filter technique applied here on the Car image, Lena image, Boat image, Houselights image, mammography image and ultrasound image. These images are corrupted by additive white Gaussian noise and multiplicative noise, like Gaussian noise. Proposed algorithm deal with noise contaminated image that is processed by block-by-block operations. A weight matrix was taken into account with suitable block size (4×4) in the proposal. Block-adaptation phase can be make a important phenomenon in the digital image processing. Proposed method implies the quality results in terms of PSNR and minimize the RMSE and visual appearance of the final image, given proposal achieved the higher PSNR, minimize RMSE and visual appeal of the final image.

KeywordsDesign of A Navel adaptive TDBLMS-Based wiener filter parallel to TDBLMS algorithm, PSNR, RMSE, Gaussian noise, Block-by-block, adaptive, weight training phase.

I. INTRODUCTION

A navel Adaptive TDBLMD-Based wiener filter provides less complicity than conventional algorithms TDBLMS. This joint process is much superior to TDBLMS algorithm for Image Noise Cancellation..

In 1981, Clark [7] extended the block processing scheme proposed by Burros [8] and proposed block the block least- mean-square (BLMS) approach. Computational complexity is dramatically reduced and provides quality of image in that approach. Besides, either parallel processing or fast Fourier transform (FFT) can be applied to accomplish the linear operations. On the other hands, the adaptive algorithms with two dimensions (2-D) are generally applied to the applications of digital image processing. An adaptive filter uses the initial weight matrix decision mechanism with the smaller block size of 4 x 4 instead of the larger ones like those in the block- adapting phase for finding a suitable weight (coefficient) matrix of the digital filter in advance. Then, treat this weight as the initial weight matrix for the processing of noise cancellation.

TDBLMS ALGORITHM

An image signal of 2-D is usually partitioned into block with a dimension of L X L for each in the 2-D disjoint block-by block image processing. An image with R rows of pixel and C columns of pixel partitioned into X block is illustrated in

Fig. 1.2 The relationship between the block index s and the

spatial block index (r, c) is [12]

S = (r-1)C/L +c (1)

Where r =1,.,R/L and C= 1,.,C/L . For convenient, the (, )-th element ((, )) of the image can be treated as the () element in the S th block and denoted as the element (,). the relationship is

+

(,) = [( 1) + , 1 ] (2) Where = 1 and = 1

The image is processed block-by-block sequentially from left to right and from top to bottom in which each pixel is convolved the pixel in a filter window with a dimension of MXN. Fig. 2 illustrates this approach which performs the operations from (3) to (5) iteratively [10]. That

is (,) =

Figure1.2. 2-D partition diagram

() = (, )[ 1] + + ( 1)

=1 =1

, ( 1) + + ( 1)

] (3)

Where (,) the image of the S-block is after processing, (, ) is the (, ) th element in the weight matrix of the S-block. The error signal (,) is the difference between the image () and the primary input image (). That is

(,) = () () (4)

The updating mechanism of the weight matrix Ws=1 of the (S+1) -th block is expressed as

large. The termination criterion (P=-10) for this face is defined as BNCR < P (6)

Where p is the termination parameter and BSNR stands for the block-noise-cancellation ratio that is define as (7)

BSNR= log10

2 ( , ) ( (

+ ,)) (5)

Where µ is the convergence factor.

II PROPOSED ADAPTIVE FILTER ALGORITHMS

The operations of this proposed adaptive filter can be divided into two phases. In beginning, the adaptive filter operates in the initial weight matrix decision phase where the initial weigh matrix for a better performance will be obtained. Then, the adaptive filter enters the block adapting phase where the TDBLMS-based wiener filter and TDBLMS-Based are algorithm parallel to applied to block-by-block process for enhancing the PSNR and minimize the RMSE for the noise image. Fig.1.3. show the block diagram of the proposed adaptive filter.

1 Initial Weight Matrix Decision Phase:

In the initial weight matrix decision phase, a suitable weight matrix will be found to be treated as the initial weight matrix W1 for the processing in the block adapting phase. First each element of initial weight matrix [1] is set a value of zero. That is WT1 = [1[, )] where the element

1[, = 0] for = 1, and = 1, Then Design of A Navel Adaptive TDBLMSBased Wiener parallel to TDBLMS algorithms for Image Noise Cancellation applied to process the original noise image in the manner of the scanning block-by-block from left to right and top to down for updating the weigh matrix of each block iteratively until the termination

criterion is reached [10]. In this phase, the block size

Figure.1.3 Proposed Design of A Navel Adaptive TDBLMSBased Wiener parallel to TDBLMS algorithms for Image Noise Cancellation with noise dependent block mechanism.

In the (7), 2 stand for the power of the reference signal ,

Rs(rb,cb) and can be expressed as

+1 1[(,)]2

2 =

=1

=1

(8)

is chosen as 4 x 4 which is smaller than in most cases (8×8, 16×16, 32×32,) and such that there are enough block for updating the weight matrix especially when the value of L is

[+1][ 1]

The term 2 is the power of the primary input signal

(,), and can be expressed as

Lt+M-1 Lt-M-1[ds(rb,cb)- X ]2

performance of parallel TDBLMS-Based wiener filter and TDBLMD overcome this problem. Moreover performance factor listed in Table 1, Table 2, and Table 3, the PSNR of the Design of A Navel Adaptive TDBLMSBased Wiener parallel to

TDBLMS algorithms for Image Noise Cancellation. RSNR become

x

2 =

K=1

L=1

mean

(9)

greater and RMSE minimizes of Design of A Navel Adaptive TDBLMSBased Wiener parallel to TDBLMS algorithms for Image

The term 2 is the power of the primary input signal

es(rb,cb), and can be expressed as

Noise Cancellation..

Lt+M-1 Lt-M-1[es(rb,cb)- X ]2

x

2 =

K=1

L=1

mean

(10)

, and stand for the means of ,

, , respectively.

2. Block Adaptive Phase:

Once the suitable weight matrix is found, then the output of the weight training phase is treated as the initial (W1) input for the block-adaptive phase.

In this phase, the sutable block is chosen that truly depends on noise contamination. After that if noise is high then take large size of block and apply TDBLMS-Based wiener filtering for the image noise cancellation else noise is low take small size of block only apply TDBLMS algorithm.

The PSNR of TDBLMS-Based Wiener filter is much better than conventional TDBLMS method in for different block size and different noise level.

The RMSE of TDBLMS-Based Wiener filter is much less than conventional method for different block size and different noise level.

III SIMULATION RESULTS

We created the primary input signal with a dimension of 256×256 in the simulation by adding a white-Gaussians noise with zero mean to the ideal image Lena, Car, and Boat, with 256 gray-levels in Fig. (a). the noise primary input image with an SNR of 0 dB shown in Fig.2.2. in the simulation convergence factor µ in (5) set to 4.5×10-7. The 4-th order transversal FIR filter is chosen to convolve the reference image. The dimension of the filter window is chosen as 4×4 (M=2, N=2). We applied four difference block of 8×8(L=8), 16×16(L=16), and 32×32(L=32), 64×64(L=64) in the

simulation for observing the effect of block size on the performance. Table 1.1 lists the performance comparison. the Design of A Navel Adaptive TDBLMSBased Wiener parallel to TDBLMS algorithms for Image Noise Cancellation using a block size of 4×4(L=4) Fig 2.4 is the restored image for the proposed adaptive filter where the termination parameter p is chosen be

-10 dB. Fig 5(1.4) show the simulation result for the block size of 4×4. It is obviously that the proposed approach cancels the noise with a nearly constant BSNR, however the performance of the TDBLMS algorithms is not good for the first several blocks. But in this proposed algorithm

1. Plot of Boats image

2. Plot of Lena image

3. Plot of Car image

Original image

1. original image

   noisy image

2. noisy image

   Weight Training phase

3. weight tarining phase

   denoised image

4. de-noised image

Figure.1.4. (a) original image (b) noisy image (c) weight tarining phase and (d) de-noised image

Table-1.1 of Boats image

Table-1.2 of Lena image

Table-1.3 of car image

IV CONCLUSION

In this work we are proposing design of a navel adaptive TDBLMS-Based Wiener parallel to TDBLMD-Based filter for image noise cancellation with noise dependent block mechanism. In this mechanism a suitable weight matrix was found by scanning the image then first, a suitable weight matrix was found by scanning the image block-by-block and updating the weight matrix for each unit the termination criterion is reached in the weight-training phase (WTP) then, the suitable weight matrix in the block adaptive phase. The simulation performed on the noise image Lena, Car, and Boats with a dimension of (256× 256) with an SNR of 0 dB shows that this approach can achieve the PSNRS values and RMSE values of Lena image, Car image, and Boat image. All the PSNR values and RMSE values result has been shown by table 1, table 2, table 3 table respectively. Above all the discussion Design of proposal provides better performance over only TDBLMS-based algorithms.

V REFERENCES

1. S. Haykin, Adaptive Filter Theory, 4th ed. Englewood Cliffs, NJ: Prentice-Hall, 2001.

2. B. Widrow, J. McCool, M. Larimore, and C. Johnson, "Stationary and nonstationary learning characteristics of the LMS adaptive filter", IEEE Proc., vol. 64, pp. 1151-1162, Aug. 1976.

3. N. A. M. Verhoeckx et aI., "Digital echo cancellation for baseband data transmission", IEEE Trans. Acoust., Speech, Signal Processing, vol. ASSP-27, pp. 768-781, Dec. 1979.

4. B. Widrow et aI., "Adaptive noise cancelling: Principles and applica tions", IEEE Proc., vol. 63, pp. 1692-1716, Dec. 1975.

5. B. Friedlander, "System identification techniques for adaptive noise canceling", IEEE Trans. Acoust., Speech, Signal Processing, vol. ASSP- 30, pp. 699-709, Oct. 1982.

6. E. H. Satorius and S. T. Alexander, "Channel equalization using adaptive lattice algorithms", IEEE Trans. Commun., vol. COMM-27, pp. 899- 905, June 1979.

7. G. A. Clark, S. K. Mitra, and S. R. Parker, "Block implementation of adaptive digital filters", IEEE Trans. Circuits Syst., vol. CAS-28, pp. 584-592, June 1981.

8. C. S. Burrus, "Block implementation of digital filters", IEEE Trans. Circuits Theory, vol. CT-18, pp. 697-701, Nov, 1971.

9. M. M. Hadhoud and D. W. Thomas, "The two-dimensional adaptive LMS (TDLMS) algorithm", IEEE Trans. Circuits Syst.,vol. 35, pp. 485- 494, May 1988.

10. W. B. Mikhael and S. M. Ghosh, "Two-dimensional block adaptive filtering algorithms", in Proc. IEEE Int. Symp. Circuits Syst., San Diego, CA, May 1992, pp. 1219-1222. 674

11. W. B. Mikhael and F. H. Wu, "A fast block FIR adaptive digital filtering algorithm with individualadaptation of parameters", IEEE Trans. Circuits Syst., vol. 36, pp. 1-10, Jan. 1989.

12. T. Wang and C. L. Wang, "A new two-dimensional block adaptive FIR filtering algorithm and its application to image restoration", IEEE Trans. Image Proc., vol. 7, pp. 238-246, Feb, 1998.