فیلتر ذره‌ای با مدل مشاهده مبتنی بر فیلتر وفقی کرنلی
محورهای موضوعی : مهندسی برق و کامپیوترحمیده حائری 1 , هادی صدوقی یزدی 2
1 - مؤسسه آموزش عالی اقبال لاهوری مشهد
2 - دانشگاه فردوسی مشهد
کلید واژه: فیلتر ذره‌ای حداقل میانگین مربعات خطای کرنلی (KLMS) حداقل مربعات بازگشتی کرنلی (KRLS) تخمین مدل,
چکیده مقاله :
هرچند که فیلتر ذره‌ای ابزاری مؤثر در ردیابی شیء می‌باشد اما یکی از محدودیت‌های موجود، نیاز به وجود مدلی دقیق برای حالت سیستم و مشاهدات است. بنابراین یکی از زمینه‌های مورد علاقه محققین تخمین تابع مشاهده با توجه به داده‌های یادگیری است. تابع مشاهده ممکن است خطی یا غیر خطی در نظر گرفته شود. روش‌های موجود در تخمین تابع مشاهده با مشکلاتی مواجه هستند و از جمله این مشکلات، وابستگی به مقدار اولیه پارامترها در روش‌های دومرحله‌ای مبتنی بر ماکسیمم‌سازی انتظار و نیازمندی به یک سری مدل از پیش تعریف شده در روش‌های مبتنی بر چند مدل می‌باشد. در این مقاله، یک روش بدون راهنما برای غلبه بر این مشکلات با استفاده از فیلترهای وفقی کرنلی ارائه شده است. به این منظور از فیلترهای وفقی حداقل میانگین مربعات خطای کرنلی یا حداقل مربعات بازگشتی کرنلی برای تخمین تابع غیر خطی مشاهده استفاده می‌شود. با فرض معلوم‌بودن تابع فرایند و با داشتن دنباله‌ای از مشاهدات، تابع مشاهده مجهول تخمین زده می‌شود. ضمناً برای کاهش هزینه محاسباتی و افزایش سرعت اجرا، از روش تُنُک‌سازی داده‌ها با استفاده از روش وابستگی خطی تقریبی استفاده شده و الگوریتم پیشنهادی در دو کاربرد مورد ارزیابی قرار گرفته است. آزمایش اول بر پیش‌بینی سری‌های زمانی و دیگری روی ردیابی اشیا در ویدئو می‌باشد. نتایج به دست آمده حاکی از برتری روش پیشنهادی در مقایسه با چند روش موجود است.الگوریتم RRT به دست آمده و مقایسه گردید. این نتایج نشانگر کارایی مناسب رویکرد پیشنهادی است.
Particle filter is an effective tool for the object tracking problem. However, obtaining an accurate model for the system state and the observations is an essential requirement. Therefore, one of the areas of interest for the researchers is estimating the observation function according to the learning data. The observation function can be considered linear or nonlinear. The existing methods for estimating the observation function are faced some problems such as: 1) dependency to the initial value of parameters in expectation-maximization based methods and 2) requiring a set of predefined models for the multiple models based methods. In this paper, a new unsupervised method based on the kernel adaptive filters is presented to overcome the above mentioned problems. To do so, least mean squares/ recursive least squares adaptive filters are used to estimate the nonlinear observation function. Here, given the known process function and a sequence of observations, the unknown observation function is estimated. Moreover, to accelerate the algorithm and reduce the computational costs, a sparsification method based on approximate linear dependency is used. The proposed method is evaluated in two applications: time series forecasting and tracking objects in video. Results demonstrate the superiority of the proposed method compared with the existing algorithms.
[1] H. Liu and F. Sun, "Efficient visual tracking using particle filter with incremental likelihood calculation," Information Sciences, vol. 195, pp. 141-153, Jul. 2012.
[2] G. G. Rigatos, "Nonlinear kalman filters and particle filters for integrated navigation of unmanned aerial vehicles," Robotics and Autonomous Systems, vol. 60, no. 7, pp. 978-995, Jul. 2012.
[3] S. Park and S. Choi, "A constrained sequential EM algorithm for speech enhancement," Neural Networks, vol. 21, no. 9, pp. 1401-1409, Nov. 2008.
[4] A. Ahmad and P. Lima, "Multi-robot cooperative spherical-object tracking in 3D space based on particle filters," Robotics and Autonomous Systems, vol. 61, no. 10, pp. 1084-1093, Oct. 2013.
[5] L. E. H. Olivier, B. Craig, and I. K Craig., "Dual particle filters for state and parameter estimation with application to a run-of-mine ore mill," J. of Process Control, vol. 22, pp. 710-717, 2012.
[6] X. Yang, K. Xing, K. Shi, and Q. Pan, "Joint state and parameter estimation in particle filtering and stochastic optimization," J. of Control Theory and Applications, vol. 6, no. 2, pp. 215-220, May 2008.
[7] Z. Messaoudi, A. Ouldali, and M. Oussalah, "Comparison of interactive multiple model particle filter and interactive multiple model unscented particle filter for tracking multiple manoeuvring targets in sensors array," in Proc IEEE 9th Int. Conf. on, Cybernetic Intelligent Systems, CIS'10, 6 pp., 1-2 Sep. 2010.
[8] X. Wang and C. Z. Han, "A multiple model particle filter for maneuvering target tracking based on composite sampling," Acta Automatica Sinica, vol. 39, no. 7, pp. 1152-1156, 2013.
[9] S. Seifzadeh, B. Khaleghi, and F. Karray, "Distributed soft-data-constrained multi-model particle filter," IEEE Trans. on Cybernetics, vol. 45, no. 3, pp. 384-394, Mar. 2015.
[10] B. V. Srinivasan, Gaussian Process Regression for Model Estimation, Electrical Engineering, University of Maryland, 2008.
[11] P. Zhu, B. Chen, and J. C. Principe, "A novel extended kernel recursive least squares algorithm," Neural Networks, vol. 32, pp. 349-357, Aug. 2012.
[12] B. Ristic, S. Arulampalam, and N. Gordon, Beyond the Kalman Filter: Particle Filters for Tracking Applications, Artech House, 2004.
[13] B. Oreshkin and M. Coates, "Bootstrapping particle filters using kernel recursive least squares," in Proc. IEEE Aerospace Conf., 7 pp., 2007.
[14] J. Yu, Y. Tang, J. Xu, and Q. Zhao, "Research on particle filter based on an improved hybrid proposal distribution with adaptive parameter optimization," in Proc. Fifth Int. Conf. on Intelligent Computation Technology and Automation, ICICTA'12, pp. 406-409, 12-14 Jan. 2012.
[15] S. Kim and J. S. Park, "Sequential Monte Carlo filters for abruptly changing state estimation," Probabilistic Engineering Mechanics, vol. 26, no. 2, pp. 194-201, 2011.
[16] N. J. Gordon, D. J. Salmond, and A. F. M. Smith, "Novel approach to nonlinear/non-Gaussian Bayesian state estimation," IEE Proc. F: Radar and Signal Processing, vol. 140, no. 2, pp. 107-113, Apr. 1993.
[17] W. L. Jose C. Principle, Simon Haykin, Kernel Adaptive Filtering: A Comprehensive Introduction, Hoboken, New Jersey: John Wiley & Sons, 2010.
[18] S. Haykin, Adaptive Filter Theory, Printice Hall, 1996.
[19] S. B. Chitralekha, J. Prakash, H. Raghavan, R. B. Gopaluni, and S. L. Shah, "A comparison of simultaneous state and parameter estimation schemes for a continuous fermentor reactor," J. of Process Control, vol. 20, no. 8, pp. 934-943, Sep. 2010.
[20] I. P. Weifeng Liu and J. C. Principe, "An information theoretic approach of designing sparse kernel adaptive filters," IEEE Trans. on Neural Networks,vol. 20, no. 12, pp. 1950-1961, Dec. 2009.
[21] D. Nguyen-Tuong and J. Peters, "Incremental online sparsification for model learning in real-time robot control," Neurocomputing, vol. 74, no. 11, pp. 1859-1867, May 2011.
[22] A. Doucet, S. Godsill, and C. Andrieu, "On sequential Monte Carlo sampling methods for Bayesian filtering," Statistics and Computing, vol. 10, no. 3, pp. 197-208, Jul. 2000.
[23] F. Desbouvries, Y. Petetin, and B. Ait-El-Fquih, "Direct, prediction-and smoothing-based Kalman and particle filter algorithms," Signal Processing, vol. 91, no. 8, pp. 2064-2077, Aug. 2011.
[24] A. C. Sankaranarayanan, A. Srivastava, and R. Chellappa, "Algorithmic and architectural optimizations for computationally efficient particle filtering," IEEE Trans. on Image Processing, vol. 17, no. 5, pp. 737-748, May 2008.
[25] M. S. Arulampalam,S. Maskell, N. Gordon, and T. Clapp, "A tutorial on particle filters for online nonlinear/non-gaussian bayesian tracking," IEEE Trans. on Signal Processing, vol. 50, no. 2, pp. 174-188, Feb. 2002.
[26] M. Netto, L. Gimeno, and M. Mendes, "On the optimal and suboptimal nonlinear filtering problem for discrete-time systems," IEEE Trans. on Automatic Control, vol. 23, no. 6, pp. 1062-1067, Dec. 1978.
[27] H. Zhen-Tao, L. Qi, and L. Xian-Xing, "A novel particle filter based on two stage prediction update," Energy Procedia, vol. 13, pp. 7389-7396, 2011.
[28] T. K. Amin Zia, J. P. Reilly, D. Yee, K. Punithakumar, and S. Shirani, "An EM algorithm for nonlinear state estimation with model uncertainties," IEEE Trans. on Signal Processing, vol. 56, no. 3, pp. 921-936, Mar. 2008.
[29] Y. L. G. Yang and P. Bhattacharya, "A driver fatigue recognition model based on information fusion and dynamic Bayesian network," Information Sciences, vol. 180, no. 10, pp. 1942-1954, 15 May 2010.
[30] S. Zheng, "An intensive restraint topology adaptive snake model and its application in tracking dynamic image sequence," Information Sciences, vol. 180, no. 16, pp. 2940-2959, 15 Aug. 2010.
[31] Z. Zivkovic, "Improved adaptive gaussian mixture model for background subtraction," in Proc. of the 17th Int.l Conf. on Pattern Recognition, ICPR'04, vol. 2, pp. 28-31, 2004.
[32] P. Gorur and B. Amrutur, "Speeded up gaussian mixture model algorithm for background subtraction," in Proc. of 8th IEEE International Conf. on. Advanced Video and Signal-Based Surveillance, AVSS'11, pp. 386-391, 30 Aug.-2 Sep. 2011.
