An efficient affine projection algorithm for 2-D FIR adaptive filtering and linear prediction
Introduction
Two-Dimensional (2-D) system identification has received extensive attention in a wide range of image processing and image analysis applications. Typical examples include modeling of 2-D data, filtering and smoothing of noisy images, image enhancement and edge detection, prediction and spectrum estimation of 2-D data, image restoration, stochastic texture analysis and synthesis, image segmentation, image compression, target detection, change detection in image sequences, etc. [1], [2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14].
A widely used class of 2-D models consists of linear 2-D finite impulse response (2-D-FIR) models, i.e., models that provide estimates of the underlying process using a finite support, linear combination of the 2-D data. Linear 2-D modeling is very popular due to the underlying model simplicity and the ability to obtain the optimum model parameters by solving a set of linear equations, for the case when the mean squared error (MSE) has been adopted as the measure of the estimation performance [6]. 2-D MSE parameters estimation has been considered in the past, for the case of stationary 2-D models, with known data statistics. In many applications, however, the classical 2-D MSE solution cannot be directly applied, because either the underlying data model is nonstationary, or the data statistics are not known in advance. In these cases, the optimum design is obtained by applying adaptive least-squares (LS) 2-D techniques, that operate directly on the available data set [13], [15], [16], [17].
LS 2-D filtering aims to shape an input signal so that the corresponding output matches a desired response signal . Following the standard notation, indices m, n denote the row and the column of the image, respectively. 2-D data are obtained by a particular scanning of a discrete 2-D rectangular region, consisting of an regularly spaced lattice. Raster scanning is assumed, i.e., left to right, top to bottom. In the LS formulation we assume that data records of input and the desired response and , over the range , provide our knowledge basis, and that we select the filter which minimizes a cost function over the available data record. The minimization procedure is carried out recursively, i.e., the optimum LS filter is re-estimated (or adapted) each time a new pair of measurements is collected.
Two different approaches, leading to two widely used algorithmic families, have been adopted for the adaptive estimation of the optimum LS 2-D FIR filter, on the basis of the available data set. The first one is based on a stochastic approximation of the steepest descent method and is known as the 2-D LMS family [1], [15], [18], [19], [20]. The latter is based on a stochastic approximation of the Gauss–Newton method and is known as the 2-D RLS family [2], [17], [21], [22], [23], [24], [25], [26]. Although 2-D LMS-type algorithms have small computational cost, they suffer from slow convergence rate, especially when the input signal autocorrelation matrix has a large eigenvalue spread. 2-D RLS-type algorithms do not suffer from such a drawback. They have, however, increased computational complexity. In an attempt to improve the convergence characteristics of the 2-D LMS-type algorithms, a Quasi-Newton adaptive scheme, called ‘the 2-D Affine Projection Algorithm (2-D-APA)’, has been proposed [28]. The good convergence performance of the 2-D-APA adaptive algorithm, is, however, overshadowed by the high computational complexity of the method.
In this paper, fast algorithms for the efficient implementation of the 2-D-APA scheme will be presented. The derivation of the fast 2-D-APA algorithms is based upon the spatial shift invariance property that the 2-D regressor vector possesses. The proposed have reduced computational complexity, compared to that of the original 2-D-APA algorithm. The performance of the proposed fast 2-D-APA algorithms is illustrated by means of computer simulation.
Section snippets
2-D FIR filtering
Linear system parameterization is an important class of system modeling with a wide area of applications. The most popular among the class of linear models is the one of a finite impulse response (FIR). This restriction is either intrinsic in the physical system modeling or it is imposed in order to simplify the estimation task and to reduce the computational load in real-time applications.
Let be the input of a linear 2-D FIR model, defined over a regularly spaced lattice
The 2-D APA
The 2-D APA algorithm has been proposed in [28], as a realization of 2-D adaptive FIR filters using the affine projection method. The 2-D APA algorithm of Muneyasu and Hinamoto [28] follows the design methodology of the 1-D counterpart [29]. The motivation of the application of the affine projection method to the design of 1-D, as well as 2-D adaptive FIR filters, is the improvement of the convergence performance that this method offers, compared to the LMS counterparts [35]. The major drawback
The fast 2-D affine projection algorithm
In this section, fast recursions will be derived for the efficient implementation of Eqs. (1.1) and (1.4) of Table 1, without the need for direct matrix product's computation. The proposed scheme is an extension of the 1-D single-channel FAPA algorithm [30], [32], and the 1-D multichannel FAPA algorithm [32]. The fast recursions that implement Eqs. (1.1) and (1.4) are derived by taking into account the spatial shift invariance properties of the data matrix .
The derivation of the fast
Simulations
The performance of the proposed 2-D fast affine projection algorithm is illustrated by means of computer simulations. Two important 2-D adaptive filtering applications are considered: (a) 2-D system identification in a stationary and in a nonstationary additive noise environment, and (b) 2-D adaptive noise canceling.
The system model in the 2-D system identification experiment is a strongly causal 2-D FIR system of the formIn this case, the filter support
Conclusions
A two-dimensional (2-D), fast affine projection algorithm (FAPA), for 2-D adaptive filtering and linear prediction, has been presented in this paper. The derivation of the proposed algorithm is based on the spatial shift invariance properties that the 2-D discrete time signals possess. The proposed algorithm has low computational complexity, comparable to that of the 2-D LMS algorithm. Simulation results indicate that the convergence speed and the tracking ability of the proposed scheme are
References (35)
- et al.
Adaptive algorithms for change detection in image sequence
Signal Processing
(1991) - et al.
A realization of 2-D adaptive filters using affine projection algorithm
J. Franklin Inst.
(1998) - et al.
A true order recursive algorithm for two-dimensional mean squared error linear prediction and filtering
Signal Processing
(2000) - et al.
Two-dimensional block diagonal LMS adaptive filtering
IEEE Trans. Signal Process.
(September 1994) - Y. Boutalis, S. Kollias, G. Carayannis, A fast multichannel approach to adaptive image estimation, Transactions of the...
- et al.
Multidimensional Digital Signal Processing
(1984) - et al.
Enhanced detectability of small objects in correlated clutter using an improved 2-D adaptive lattice algorithm
IEEE Trans. Image Process.
(March 1997) - A.K. Jain, Advances in mathematical models for image processing, Proc. IEEE (May 1981)...
Fundamentals of Digital Image Processing
(1989)- H. Kaufman, A.M. Tekalp, Survey of estimation techniques in image restoration, IEEE Control Systems (January 1991),...
Two-Dimensional Signal and Image Processing
Adaptive linear predictive coding of time-varying images using multidimensional recursive least squares ladder filters
IEEE J. Sel. Areas Comm.
Performance evaluation of 2-D adaptive prediction filters for detection of small objects in image data
IEEE Trans. Image Process.
Two-dimensional adaptive prediction, smoothing and filtering
IEE Proc. Pt. F
The two-dimensional adaptive LMS (TDLMS) algorithm
IEEE Trans. Circuits Systems
Cited by (8)
A novel 2D-ABC adaptive filter algorithm: A comparative study
2015, Digital Signal Processing: A Review JournalCitation Excerpt :On the other hand, this improvement is an important trade-off between the computational complexity and the performance. An improved version of the 2D APA algorithm was described by Glentis [13]. This new version reduced the computational complexity compared with the previous version and has a fast convergence speed with good observation capability.
The novel two-dimensional adaptive filter algorithms with the performance analysis
2014, Signal ProcessingCitation Excerpt :Unfortunately, this improvement comes at the expense of a higher computational complexity. In [6], a fast affine projection algorithm for two-dimensional adaptive linear filtering was presented. The results show that this algorithm has fast convergence speed and good tracking ability.
Multispectral satellite image denoising via adaptive cuckoo search-based wiener filter
2018, IEEE Transactions on Geoscience and Remote SensingTwo-Dimensional CS Adaptive FIR Wiener Filtering Algorithm for the Denoising of Satellite Images
2017, IEEE Journal of Selected Topics in Applied Earth Observations and Remote SensingImage denoising with two-dimensional adaptive filter algorithms
2011, Iranian Journal of Electrical and Electronic Engineering