Number Of Decomposition Levels Research Articles

Skew detection using wavelet decomposition and projection profile analysis

In this paper, a novel document skew detection algorithm based on wavelet decompositions and projection profile analysis is proposed. First, the skewed document images are decomposed by the wavelet transform. The matrix containing the absolute values of the horizontal sub-band coefficients, which preserves the text’s horizontal structure, is then rotated through a range of angles. A projection profile is computed at each angle, and the angle that maximizes a criterion function is regarded as the skew angle. Experimental results show that this algorithm performs well on document images of various layouts and is also robust to different languages. The effects of various wavelet basis, number of decomposition levels, and parameters of the criterion function are investigated too.

Application and optimization of the discrete wavelet transform for the detection of broken rotor bars in induction machines

The problem of the bar breakage diagnosis in electrical induction cage machines is a matter of increasing concern nowadays, due to the widely spread use of these machines in the industry. The classical approach, focused on the Fourier analysis of the steady-state current, has some drawbacks that could be avoided if a study of the transient behavior of the machine is performed. The discrete wavelet transform (DWT) is an ideal tool for this purpose, due to its suitability for the analysis of signals whose frequency spectrum is variable in time. The paper shows how the study of the high-level signals resulting from the DWT of the transient starting current of an induction motor allows the detection of a particular characteristic harmonic that occurs when a rotor bar breakage has taken place. This constitutes an alternative approach that avoids some problems that the traditional method implies and that can even lead to a wrong diagnosis of the fault. In the work, the application of the DWT for broken bar detection is optimized, regarding certain parameters of the transform such as type of the mother wavelet, number of decomposition levels, order of the mother wavelet and sampling frequency.

Performance comparison of two-dimensional discrete wavelet transform computation schedules on a VLIW digital signal processor

The two-dimensional discrete wavelet transform (2D DWT) is becoming one of the standard tools for image and video compression systems. Various input-traversal schedules have been proposed for its computation. Here, major schedules for 2D DWT computation are compared with respect to their performance on a very long instruction-word (VLIW) digital signal processor (DSP). In particular, three popular transform-production schedules are considered: the row–column, the line based and the block based. Realisations of the wavelet transform according to the considered schedules have been developed. They are parameterised with respect to filter pair, image size and number of decomposition levels. All realisations have been mapped on a VLIW DSP, as these processors currently form an attractive alternative for the realisation of signal, image and video processing systems. Performance metrics for the realisations for a complete set of parameters have been obtained and compared. The experimental results show that each realisation performs better for different points of the parameter space.

On the Use of Wavelet-Based Denoising to Improve Power Delay Profile Estimates from 1.8 GHz Indoor Wideband Measurements

This work presents an alternative processing scheme to improve measurement-based power delay profiles (PDP) estimates, using wavelet-based denoising. The usual PDP processing comprises cutting off all the data below a previously determined flat noise threshold. The proposed method is a more refined approach, since it tries to extract the noise from the measured data, keeping only the estimated signal. Wavelet denoising can be performed in many different ways, for it depends on several parameters, like the noise threshold selection rule, the wavelet function, the choice or not for a wavelet coefficients shrinkage and the number of decomposition levels. Several parameter combinations have been tested, in special for the Visu selection rule, which presented the best performance for the available data in the overall. Denoising was applied to real data from indoor environments, collected from wideband channel sounding surveys, centered at 1.8 GHz. Since frequency domain sounding has been carried out, denoising has been tested both directly over the frequency domain, and over the time--delay domain (PDP). The major result of the proposed processing was a qualitative improvement of the PDP, with smoother noise floors, and also with increases up to 8 dB on signal-to-noise ratios, in special for delay domain denoising.

Design of a low-power real-time wavelet CODEC circuit

Abstract This paper presents the architecture and design of a CODEC circuit that can compress and reconstruct 4:2:2 color VGA video images using two-dimensional DWT. For real-time operation and low-power consumption, we modified the traditional Mallat's sub-band coding method to reduce the amount of computation required in two-dimensional DWT. Using the modified method, two-dimensional DWT can be performed by two sets of one-dimensional DWT operations regardless of the number of decomposition levels. We adopted the systolic array architecture to design the circuit implementing our method. Our low-power real-time CODEC circuit was designed using Verilog HDL and fabricated as a VLSI chip using 0.35 μm 3.3 V CMOS standard cell process. We made a simple DVR system using our CODEC chip to store and retrieve the video image data to and from the hard disk, and observed successful real-time operations.

Wavelet Filtering of Signals from Ultrasonic Flaw Detector

The paper considers basic principles of the algorithm of wavelet filtering of signals from an ultrasonic aw detector with the help of the fast wavelet transform. Criteria for selection of the type and order of a wavelet are formulated, and the number of decomposition levels required for the needed accuracy and reproducibility of results of ultrasonic tests are determined.

Une méthode de décomposition de domaines multifrontale multiniveaux

ABSTRACT The developpement of multiprocessor computers has led to a new interest in domain decomposition methods. The goal of this article is to present a multilevel domain decomposition algorithm for the finite element method using a direct solver on the interface problems at each level. In order to reduce the cost of building the inverse of the rigidity matrices of each subdomain, a multifrontal method is used. After a brief description of the Schur complement method and its classical implementation, the multifrontal method is presented in detail. A comparison of theoretical number of operations and measured execution times between these two implementations is given. In the second part of the article, the influence of two parameters is studied: the number of subdomains and the number of decomposition levels. Finally, a comparison between the multifrontal domain decomposition method and a method without decomposition is provided.