A new proportionate affine projection sign algorithm is proposed for network echo cancellation. It uses a recursive procedure and takes into account the previously computed proportionate coefficients. It is shown that the proposed algorithm can obtain a lower steady-state misalignment than other affine projection sign algorithms for different echo paths, impulsive interferences and step sizes.
The approximate memory improved proportionate affine projection algorithm has been proposed for sparse system identification. This paper presents a fast recursive implementation of this algorithm. Three ideas used previously for other affine projection variants are used: auxiliary coefficients vectors, periodically update of the proportionate coefficients and recursive filtering of the error vector. Simulation results are made in order to show the performance of the algorithm for network echo cancellation example.
In this paper, a fixed-budget implementation of the kernel proportionate normalized least mean square (KPNLMS) algorithm using a variable step size scheme is proposed. The similarity between the equations of the NLMS algorithm and those of the kernel proportionate NLMS algorithm with coherence criterion is emphasized and the reason of using the proportionate coefficients for the KPNLMS algorithm is given. It is shown that applying the proportionality principle to the kernel outputs leads to better convergence properties than applying it to the weights of the nonlinear filter. The effect of the step size on the convergence properties of KPNLMS is exemplified. Also, and the effect of SNR on the dictionary size of the KPNLMS algorithm is proved for channel equalization and forward prediction examples. The influence of the dictionary size on the performance of the fixed budget KPNLMS algorithm is demonstrated. Therefore, a simple variable step size scheme is proposed in order to improve the convergence properties of fixed-budget KPNLMS algorithm for channel equalization of a multi-path Rayleigh fading channel and forward prediction applications. It is also proved that the additional computational complexity burden of the proposed algorithms is very small.
Affine projection algorithm encompasses a family of configurable algorithms designed to improve the performance of other adaptive algorithms, mainly LMS based ones, especially when input data is highly correlated. The computational cost of the affine projection algorithm depends largely on the projection order, which in turn conditions the speed of convergence, thus high speed of convergence implies usually high computational cost. Some real-time applications (especially multichannel) using the affine projection algorithm can not be implemented in the existing general-purpose hardware, because of this several improvements of the affine projection algorithm have been proposed to make it more computationally efficient and more versatile in terms of performance. This paper outlines the evolution of the affine projection algorithm and its variants, in order to get an efficient and self-reconfigurable algorithm. Furthermore new improvements over the existing low cost and variable step size and projection order versions are proposed to give examples of the new generation of affine projection algorithms.
In a previous study we proposed a modified version of the recursive leastsquare (RLS) adaptive algorithm suit able for fixedpoint implementation. Using an asymptotically unbiased estimator for the algorithm's cost function we reduced the dynamic range of this parameter. In this paper, we extend the procedure for the case of QRdecompos ition� based leastsquares lattice (QRDLSL) adaptive algor ithm, a fast member of RLS family, with good numerical properties. The reduced dynamics of the algorithm's parameters leads to facility for fixedpoint implem entation. The simulations performed on a fixedpoint digital signal processor (DSP) sustain the practical aspects of the theoretical findings.
This paper presents an efficient time-domain Generalized Sidelobe Canceller (GSC) with low signal distortion capabilities using the variable step size affine projection algorithm (VSS-APA) and a log-energy based voice activity detector (VAD). The performance of the proposed VSS-APA based GSC method with integrated log-energy VAD, is illustrated in the context of speech reinforcement application using different signals with low signal-to-noise ratio and different types of noise.
The paper presents a new technique of efficient dynamic range compression and shadow compensation for still color images. The proposed method enhances low light areas while preserving the colors and details, without generating visual artifacts. The approach is based on recursive filtering and contrast stretching techniques, driven by statistical measures of the image and implemented under a logarithmic image processing model. The implementation can be used for any image represented in the RGB or YCbCr color spaces.
It is well known that the affine projection algorithm (APA) offers a good tradeoff between convergence rate/tracking and computational complexity. Recently, the evolutionary APA (E-APA) with a variable projection order has been proposed. In this paper, we propose a variable step size (VSS) version of the E-APA, called VSS-E-APA. It is shown that the VSS-E-APA is robust to near-end signal variations. Also, it has both a fast convergence speed and a small steady-state error and a much reduced numerical complexity than the VSS-APA.
The evaluation of early school-aged children's handwritten symbols is a challenging problem. The teaching of handwriting is still an essential skill in effective written communication. There is a need for an automatic quality evaluation of handwritten symbols in order to assess the progress of children ability to write nice letters or other symbols. In our case, the letters are positioned in a well defined two-dimensional space, similar to the special notebooks children use in school when learning how to write. For this study, a Wacom system composed of a tablet and a digital pen is used for collecting the data that will be sent to the analysis and evaluation module. While the school-aged child writes a character, the pen transmits the (x,y) coordinates and the time t. Next, a normalization is needed in order to keep the same distance between neighboring pixels. Several approaches are investigated (e.g. normalization with/without interpolation). The coordinates sequence is transformed in a sequence of angles measured relatively to the X axis for each written character. They encode the changing directions during pen movement. An algorithm is used in order to detect if the overall shape of the written symbol is correct. Several parameters that characterise the written character are investigated (e.g. ,,centre of mass", height over width ratio, alignment errors etc.). Their correlation with subjective scores is verified. Several metrics are proposed based on spatial and temporal measurements. Next, the handwritten quality using the legibility, form size and alignment of the letters or digits is investigated. It is shown that a rough discrimination between proficient and non-proficient handwriting can be obtained by considering the size and space parameters. Our simulations have revealed the importance of good handwritten reference samples. The goal is to develop a calligraphic handwritting learning system designed for first grade or pre-school childrens. Further research is needed in order to address other aspects of an intelligent tutor.
In this paper, two proportionate affine projection sign algorithms are introduced. The performance of the proposed algorithms is compared with that of other proportionate affine algorithms under impulsive interference environment of a network echo cancellation system. It is shown that one of the proposed algorithms, termed memory improved proportionate affine projection sign algorithm (MIP-APSA), is the most robust to impulsive interferences and colored inputs. It is proved that MIP-APSA is a good candidate for network echo cancellation, because of its low complexity, good convergence speed and tracking abilities for echo paths with different sparseness measures, and projection orders.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)
'/%3e%3c/defs%3e%3cpath fill='%23012169' d='M0 0h10080v5040H0z'/%3e%3cpath stroke='%23fff' stroke-width='.6' d='m0 0 6 3m0-3L0 3' clip-path='url(%23b)' transform='scale(840)'/%3e%3cpath stroke='%23e4002b' stroke-width='.4' d='m0 0 6 3m0-3L0 3' clip-path='url(%23c)' transform='scale(840)'/%3e%3cpath stroke='%23fff' stroke-width='840' d='M2520 0v2520M0 1260h5040'/%3e%3cpath stroke='%23e4002b' stroke-width='504' d='M2520 0v2520M0 1260h5040'/%3e%3cg fill='%23fff'%3e%3cuse xlink:href='%23d' x='2520' y='3780'/%3e%3cuse xlink:href='%23a' x='7560' y='4200'/%3e%3cuse xlink:href='%23a' x='6300' y='2205'/%3e%3cuse xlink:href='%23a' x='7560' y='840'/%3e%3cuse xlink:href='%23a' x='8680' y='1869'/%3e%3cuse xlink:href='%23e' x='8064' y='2730'/%3e%3c/g%3e%3c/svg%3e)