A Fuzzy-Based Impulse Noise Detection and Cancellation for Real-Time Processing in Video Receivers

Transcription

1 780 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 52, NO. 3, JUNE 2003 A Fuzzy-Based Impulse Noise Detection and Cancellation for Real-Time Processing in Video Receivers Chung-Bin Wu, Bin-Da Liu, Senior Member, IEEE, and Jar-Ferr Yang, Senior Member, IEEE Abstract In this paper, a simple fuzzy-based algorithm to remove the impulse noise from images is proposed. To achieve real-time applications, the proposed filter architecture, which combines fuzzy noise detection and noise filtering, is also designed. With low computational complexity, simulation results show that the proposed filters effectively remove the impulse noise. Index Terms Fuzzy, impulse noise, noise detection, noise filter, real-time hardware design. I. INTRODUCTION THERE are two types of noise that are likely to contaminate images during analog transmission or storage. When the received or stored signal level is too low, the incorporated noise is mostly Gaussian type. However, the most annoying noise, which can degrade the image quality under any condition, is impulse noise. Various restoration techniques have been proposed for the removal of impulse noise. Usually, linear filters produce serious image blurring while removing the impulse noise. Generally, a good impulse noise cancellation scheme should suppress the noise while preserving the edges and the image details. With this consideration, nonlinear techniques provide more satisfactory results than linear approaches. Nonlinear methods, including median filters [1] [3], frame average or line average filters [4], rank-order filters [5], and fuzzy-logic-control filters [6] [9] with dferent computational complexities and memory requirements, can effectively remove the impulse noise and maintain sharp edges. For example, a median filter usually requires a complicated ordering circuit, especially for a large window size. Recently suggested median filters have [1], [10]. Zhang et al. [11] used a two-step digital image-processing scheme to filter out the impulse noise. In order to achieve the line map for impulse noise identication, this scheme requires one field buffer and three line-buffers. Windyga suggested a generic -dimensional filter [12], which achieves much faster than the median filter, while performing comparably in terms of both image information conservation and noise reduction. This filter should either eliminate or attenuate most noisy pixels in synthetic and natural images not excessively contaminated. Manuscript received October 28, 2000; revised March 3, This work was supported by the National Science Council of Taiwan, R.O.C., under Grant NSC E The authors are with the Department of Electrical Engineering, National Cheng Kung University, Tainan, Taiwan, R.O.C. ( bdliu@spic.ee. ncku.edu.tw). Digital Object Identier /TIM Fuzzy set theory methods with the capability to efficiently represent the input/output relationships of dynamic systems have gained attention, especially in pattern recognition and computer vision applications [13]. Hence, there have been many efforts to develop a fuzzy filter for signal/image processing. Russo [6] and Farzam [8] proposed fuzzy logic filters to receive promising results for image enhancement. In [6] and [8], fuzzy rules were used for detecting noise and preserving images. However, most of the fuzzy logic filters are inefficient and time-consuming in real time applications. In the TV receiver set, which receives transmitted analog images, a trade-off must be made among its hardware cost, operating speed, and image quality. Rule-based fuzzy filters, which use large rulebases to process highly corrupted images, typically have high performance but demand high computational complexity. In other words, for low probability noisy images, it is obvious that only a few rules are necessary. In real-time applications, however, a limited performance in removing noise is the cost for low computational complexity. In this paper, we extend our previous work [14] and propose a new nonlinear filter, based on a simple fuzzy-based detection, to effectively maintain image quality upon the removal of impulse noises. In Section II, we propose a fuzzy-logic noise detection and threshold filter. Simulation results show that this noise detection system performs better in removing impulse noises than the traditional median filter, which requires complex hardware. In Section III, the architecture for the proposed filter is designed. A short summary of this paper is provided in Section IV. II. FUZZY-BASED AVERAGE AND THRESHOLD FILTER The intensity of an impulse noise impaired pixel is usually much larger (or smaller) than that in the surrounding pixels. When a small window is centered at an impulse-impaired pixel, the effect of the impulse noise is observed after an average filter is applied. With this concept in mind, we propose a fuzzy based nonlinear filter to reduce the impulse noise based on an block-processing unit. Fig. 1 shows a 3 3 processing block in which the corresponding eight pixels are labeled from to and the center pixel is labeled. In order to avoid reducing the resolution of TV pictures in the direct replacement, which might possess the same disadvantage as most average filters, a fuzzy noise-detecting unit was adopted. The ideal detail-preserving processor for impulse noise removal is able to detect a noise pixel by estimating the disparity in the original gray level of a noise pixel from the averaged gray /03$ IEEE

2 WU et al.: FUZZY-BASED IMPULSE NOISE DETECTION 781 Fig. 1. Nine-element processing window. (a) (b) Fig. 2. Basic block diagram for proposed filter. level of the neighboring pixels [15]. The restoration process is divided into two steps. In the first step, the pixel at the center of the block is detected by the designed noise detector. If the noise detector identies a pixel with no impulse noise, it remains unchanged. Once the noise detector locates a pixel corrupted by impulse noise, the original gray level of the noise pixel is estimated from the neighboring pixels in the second step. To detect impulse noise, a fuzzy-based average threshold (FAT) filter is proposed to restore the image. We define the following two-parameter membership function to implement fuzzy reasoning (c) (1) (d) where and are the two pre-determined parameters. For images with 256 gray levels, and, the membership function is likely to be a triangular-shaped function. is defined as the noise detection threshold, it can be justied by the type of images. From experiments, to achieve real time applications and hardware design, and as 32 and 255 are selected, respectively. Fuzzy reasoning resorts to a set of rules in order to detect noise pulses. For the proposed operator, a fuzzy set labeled noise detector (ND) is adopted to detect the noise using the membership function in (1). The noise detection algorithm is based on a simple fuzzy rule, shown as follows: (2) (e) Fig. 3. Simulation results. (a) Original image Lady ; (b) adding 5.05% impulse noise; (c) output from nine-points median filter; (d) output from fuzzy filter [6]; (e) output from FAT filter. where denotes the luminance dferences and the output. Since TV signals are transmitted in line scanning format to the receiver, the probability that impulse noise

3 782 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 52, NO. 3, JUNE 2003 TABLE I PSNRS FOR LADY IMAGE will occur simultaneously in the vertical direction is minuscule. The vertical average signal can be used as a noise detection reference. For a simple hardware design, is computed as Since factors 1/4 and 1/2 can be realized using 2-bit and 1-bit shters, rule (2) exploits the luminance dference and,, and are the inputs for (1) detecting a noise pulse in the 0 position. From (2),, and are smaller than, the pixel is detected as nonimpulse noise ( ), the pixel will not require any processing. Otherwise, the pixel in the center is detected as impaired with impulse noise, it will be replaced by a new estimated value. From [6], the output is obtained as follows: (4) where Fig. 2 depicts the detailed flow diagram of the FAT filter. In dferent image communication channels, dferent impulse noises might be generated in terms of the pulse intensity and duration. In order to achieve the optimal restoration of the impaired picture, the threshold of the impulse noise detection must be adjusted according to the type of images. Mean variance [16] can be used to define the type of image. The mean variance should have positive correlation with the threshold. The larger the mean variance is, the bigger the threshold will be selected. Nevertheless, in this paper, in order to reduce the hardware complexity, from experiments, the fixed threshold is used to implement the hardware. From (2), it can be observed that the proposed fuzzy algorithm with only one fuzzy rule is less than those suggested in [6] with 26 rules for every processed block and suitable for hardware implementation. To very the proposed filter algorithm, the Lady image with various impulse noise ratios was generated as the input for (3) (5) the FAT impulse noise filter. The noise image model was simulated by randomly adding extremely light or dark gray level pixels. At the same time, the traditional median filters with filter length of 7 and 9 were also adopted in the simulations for references. Table I shows the PSNR values achieved by the median, the peak-and-valley filter [12], the BAT filter [14], the fuzzy-based filter [6], and the proposed FAT filter. It shows that the proposed FAT filter is better than the traditional median filters and fuzzy-based filter [6] when the impulse noise ratio is less than 4.37%. With highly corrupted noise, corrupted by more than 5.05% impulse noise, the performance of the proposed method is slightly worse than the fuzzy-based filter [6] but better than the traditional median methods. In addition, the proposed FAT filters require less computation than the fuzzy filters [6], [8] which require 26 fuzzy rules or more for every processed pixel. When the ratio of impulse noise is 5.05%, Fig. 3(a), (b), (c), (d), and (e) show the original, the impulse-noise impaired, the nine-point median filtered the fuzzy-based filter [6] and the FAT filtered Lady images, respectively. The proposed FAT filter, which has less computation and achieves better performance than the median filters and fuzzy-based filter [6], should be a good impulse noise removal. III. REALIZATION OF FAT FILTER In this section, the hardware design for the FAT filter implementation is presented. As described in Fig. 2, the design procedure consists of preprocessing, noise detection and noise filtering. The preprocessing is used to compute, for each 3 3 block, is computed by (3) as described in Section II. Noise detection uses fuzzy rule in (2) and membership function in (1) to detect impulsive noise. From (2), the pixel is detected as nonimpulse noise, the pixel will not require any processing. Otherwise, the pixel in the center is detected as impaired with impulse noise, it will be replaced by a new estimated value; noise filtering processes the center pixel when the noise is detected and replaces the center pixel by the output as described in (4) and (5). The detail hardware design block diagrams are described as follows. Fig. 4 demonstrates the architecture of the proposed FAT filter. For a 3 3 processing unit, two line buffers are needed after the analog video signal is sampled by the A/D converter.

4 WU et al.: FUZZY-BASED IMPULSE NOISE DETECTION 783 Fig. 4. Proposed architecture of the FAT filter kernel. (a) Fig. 6. FAT filter processing kernel. (b) Fig. 5. (a) Image data and processing windows. (b) Operation of the data flow. Fig. 5(a) shows the relation among the three image data processed lines as the input for the noise reduction kernel depicted in Fig. 6, where,, and denote the image pixels that come from the A/D converter and the two line buffers, respectively. Since every window contains nine points, the average value, can be computed in each cycle. In the first and third clock cycles, the summations of the three column pixels are calculated. However, in the second clock cycle, only two pixels are taken into account. Since a 3 3 sample window is adopted, a pipeline architecture is established to improve the processing speed, as shown in Fig. 5(b). In this architecture, only two adders, one 1-bit shter for dividing by 2 and one 2-bit shter for dividing by 4 are required. Fig. 7 shows the function blocks for the FAT filter, which involve noise detection and noise filtering. Since the pipeline structure was used in the previous stage, we can obtain SA for every clock cycle and calculate the dference value easily Fig. 7. Fuzzy-based noise detection and filtering. using an abstract subtractor. An accumulative comparator is used to find the maximum value in Fig. 7. If the current value is larger than the held value, it will replace the current value and store it; otherwise, the maximum value will be held. The function block is used to calculate the function of as presented in the previous section. The function block is regarded as a filtering output. If the noise is detected, the output will be replaced by, otherwise, the current pixel will be the output. To design easily, in this hardware implementation, the closest neighboring pixel to the current vertical average value in the processing block is selected as output.

5 784 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 52, NO. 3, JUNE 2003 IV. CONCLUSION In order to improve the image and simply the computational complexity, a FAT nonlinear filter was proposed for traditional analog video receivers. To evaluate the proposed algorithm, various percentages of impulse noise was added to the original images. Simulation results have shown that the proposed FAT filter achieves better performance in comparison to the well-known median filters. The proposed filter requires lower computational complexity than the later in hardware. A corresponding architecture for impulse noise cancellation has also been proposed. REFERENCES [1] J. Siu, J. Li, and S. Luthi, A real time 2-D median based filter for video signals, IEEE Trans. Consumer Electron., vol. 39, pp , May [2] C. K. Kuo, C. H. Lin, and J. H. Leu, Noise reduction and concealment for block encoder images, IEEE Trans. Consumer Electron., vol. 40, pp , Nov [3] T. Kaspairs and J. Lane, Adaptive scratch noise filtering, IEEE Trans. Consumer Electron., vol. 39, pp , Nov [4] S. Inamori, S. Yamauchi, and K. Fukuhara, A method of noise reduction on image processing, IEEE Trans. Consumer Electron., vol. 39, pp , Nov [5] R. C. Hardie and G. R. Arce, Ranking in RP and its use in multivariate image estimation, IEEE Trans. Circuits Syst. Video Technol., vol. 1, pp , June [6] F. Russo and G. Ramponi, A fuzzy filter for image corrupted by image noise, IEEE Signal Processing Lett., vol. 3, pp , June [7] F. Russo, Edge detection in noisy images using fuzzy reasoning, IEEE Trans. Instrum. Meas., vol. 47, pp , Oct [8] F. Farbiz, M. B. Menhaj, S. A. Motamedi, and M. T. Hagan, A new fuzzy logic filter for image enhancement, IEEE Trans. Syst. Man, Cybern. B, vol. 30, pp , Feb [9] Y. Choi and R. Krishnapuram, A robust approach to image enhancement based on fuzzy logic, IEEE Trans. Image Processing, vol. 6, pp , June [10] S. B. Lee, A. Ortiz, R. F. Lepard, S. R. Shaw, and J. L. Kirtley, Applications of real-time median filtering with fast digital and analog sorters, IEEE/ASAME Trans. Mechatron., vol. 2, pp , June [11] Q. Zhang and R. Ward, Automatic assessment of signal-to-thermal noise of television images, IEEE Trans. Consumer Electron., vol. 41, pp , Feb [12] P. S. Windyga, Fast impulsive noise removal, IEEE Trans. Image Processing, vol. 10, pp , Jan [13] J. S. Kim and H. S. Cho, A fuzzy logic and neural network approach to boundary detection for noisy imagery, Fuzzy Sets Syst., vol. 65, pp , Aug [14] C. B. Wu, B. D. Liu, J. F. Yang, and S. C. Hsia, Algorithm of digital impulse noise cancellation, in Proc. IEEE APCCAS, Chiangmai, Thailand, Nov. 1998, pp [15] T. H. Yu and S. K. Mitra, Detail-preserving impulse noise removal of images using modied dynamic programming, IEEE Trans. Acoust., Speech, Signal Processing, vol. 4, pp , Apr [16] A. V. Oppenhenim and R. W. Schafer, Discrete-Time Signal Processing. Englewood Clfs, NJ: Prentice-Hall, Chung-Bin Wu was born in Kaohsiung, Taiwan, R.O.C., on June 22, He received the B. S. and M. S. degrees in electrical engineering from the National Cheng Kung University, Tainan, Taiwan, in 1995 and 1997, respectively. He is currently pursuing the Ph.D. degree. His research interests include VLSI design, video compression, and digital signal processing. Bin-Da Liu (S 79 M 82 SM 95) received the B.S., M.S., and Ph.D. degrees in electrical engineering from the National Cheng Kung University (NCKU), Tainan, Taiwan, R.O.C., in 1973, 1975, and 1983, respectively. From 1975 to 1977, he served as Electrical Officer in the Combined Service Forces. Since 1977, he has been on the faculty of the NCKU, where he is currently Professor in the Department of Electrical Engineering. From 1983 to 1984, he was a Visiting Assistant Professor in the Department of Computer Science, University of Illinois, Urbana-Champaign. From 1988 to 1992, he was the Director of Electrical Laboratories, NCKU, where he was the Associate Chairman of the Electrical Engineering Department from 1996 to 1999 and the Chairman from 1999 to From 1990 to 1993, he was a member of the Evaluation Committee for Junior Engineering College, Ministry of Education. Since 1995, he has been a Consultant of the Chip Implementation Center, National Science Council. Since 2002, he has been the Coordinator of the Digital IP Consortium, VLSI Educational Program of Ministry of Education, and the Coordinator of the SoC Design Promotion Program of National Science Council. He has published more than 170 technical papers. He also contributed to chapters in the book Neural Networks and Systolic Array Design (D. Zhang, Ed., Singapore: World Scientic, 2002) and the book Accuracy Improvements in Linguistic Fuzzy Modeling (J. Casillas, O. Cordon, F. Herrera, and L. Magdalena, Eds., Heidelberg, Germany: Springer-Verlag, 2003). His current research interests include physical design and testing for VLSI circuits, SoC system integration and verication, and VLSI implementation for fuzzy-neural networks and video signal processors. Dr. Liu is a member of Phi Tau Phi, Taiwan SOC Consortium, International Union of Radio Science, Chinese Fuzzy Systems Association, Chinese Institute of Electrical Engineers, and Taiwan Integrated Circuit Design Society. He has been the Chairman of IEEE Circuits and Systems Society, Taipei Chapter, since He received the Dragon Distinguished Paper Award from the Acer Foundation in 1991 and 1997, the Best Paper Award from the CIEE in 1995and 2002, the Golden Silicon Award from the Macronix Foundation in 2001 and 2002, MPC Chip Design Award from the CIC in 2002, and the Research Award from the National Science Council annually since He organized the Taiwan Student VLSI Design Contest in 1998, 1999, and Since 1992 he has served as a member of the Steering Committee of VLSI Design/ CAD Symposium and served as the General Chair in Since 1997, he has served as a member of the Technical Program Committee of R.O.C. National Conference on Fuzzy Theory and Its Applications. He served as a member of the Program Committee of the 1998 and 1999 IEEE Workshop on VLSI Signal Processing Systems. He also served as a member of the Technical Program Committee of the 1998 and 2000 IEEE Asia Pacic Conference on Circuits and Systems, and the First, Second, and Third IEEE Asia Pacic Conference on ASICs. He has been a member of the International Steering Committee of the IEEE Asia-Pacic Conference on Circuits and Systems since He was the General Co-Chair of the First International Meeting on Microsensors and Microsystems in 2003 and will be the General Chair of the 2004 IEEE Asia-Pacic Conference on Circuits and Systems. Jar-Ferr Yang (M 88 SM 98) was born in Keelung, Taiwan, R.O.C., on September 15, He received the B.S. degree from the Chung-Yuan Christian University, Taiwan, in 1977, the M.S. degree from the National Taiwan University, Taiwan, in 1979, and the Ph.D. degree from the University of Minnesota, Minneapolis, in 1988, all in electrical engineering. He was an Instructor in the Chinese Naval Engineering School from 1979 to He was with the Data Transmission and Network Design Research Group, Telecommunication Laboratories, from 1981 to He was the Chairman of the Center for Computer and Communication Research of the National Cheng Kung University (NCKU), Tainan, Taiwan, from 1997 to He was a Visiting Scholar at the Department of Electrical Engineering, University of Washington, Seattle, in Currently, he is the Deputy Director of the Electrical and Information Technology Center and a Professor at the Department of Electrical Engineering, NCKU. His teaching and research are primarily in the areas of signal processing, video, audio, and speech compression, spectrum estimation, and neural networks.