Narrow Results

In this paper, an off-line double density discrete wavelet transform based de-noising and baseline wandering removal methods are proposed. Different levels decomposition is used depending upon the noise level, so as to give a better result. When the noise level is low, three levels decomposition is used. When the noise level is medium, four levels decomposition is used. When the noise level is high, five levels decomposition is used. Soft threshold technique is applied to each set of wavelet detail coefficients with different noise level. Donoho's estimator is used as a threshold for each set of wavelet detail coefficients. The results are compared with other classical filters and improvement of signal to noise ratio is discussed. Using the proposed method the output signal to noise ratio is 19.7628 dB for an input signal to noise ratio of -7.11 dB. This is much higher than other methods available in the literature. Baseline wandering removal is done by using double density discrete wavelet approximation coefficients of the whole signal. This is an unsupervised method allowing the process to be used in off-line automatic analysis of electrocardiogram. The results are more accurate than other methods with less effort.

Based on completely different properties of the signal and noise in wavelet transform, the noise confidence factor is introduced to estimate the clutter power level by means of CM (Censored Method). A radar signal CFAR (Constant False Alarm Rate) detection method with soft threshold is proposed. The CFAR characteristic of the method under background with Gaussian distribution clutter is studied theoretically. And the experiment results of radar signal processing demonstrate that this method can detect the targets in other different clutters effectively, which shows the method's robustness and effectiveness.

As we know, a challenge of image denoising is how to preserve the edges of an image when reducing noise. In this paper, by showing the model of noisy images and taking advantage of the multiresolution analysis with wavelet transform to remove the noise, we propose a wavelet image thresholding scheme. The size of the threshold is interrelated with the noise degree, and then we present an efficient denoising method. Experimental results demonstrated that this algorithm could achieve both good visual quality and high SNR for the denoised images.