Use of a trained denoising autoencoder to estimate the noise level in the ECG

Noise level estimation plays an important role in many applications of signal and image processing, like denoising, compression and detection. Recently, deep neural networks have also been increasingly used for this purpose. In this paper, we develop an effective algorithm of noise level estimation of ECG segments based on trained denoising autoencoder (DAE) with a statistical thresholding method. An important observation is that a well-trained DAE model provides a clean representation of the corrupted training dataset. Two identical cascaded trained DAE models are considered to estimate the statistical properties, e.g., mean and standard deviation, from the trained DAE outputs after applying noise free aligned and jittered training dataset respectively. Two statistical thresholds are calculated from these statistical properties to classify whether the ECG segment is noise-free or jittered or noisy segment. The accuracy of the proposed method is quite promising in classifying and estimating unknow noise level.

Methods for image noise level estimation

The article presents a method for estimating the level of discrete white noise in an image, based on the use of linear difference operators with a vector mask. Two variants of a new method for estimating the noise level are proposed, which differ in the accuracy of the obtained estimates and computational complexity. The first version of the method can be attributed to the class of block methods, whereas the second one is intended for the rapid image analysis and is based on processing a small number of rows or columns of an image.

Transform and statistical parameter-based image noise level prediction

Noise level estimation in an image is important and useful in many image processing algorithms such as image denoising, image segmentation, and image compression. Accurately estimating the noise level without prior knowledge of the image is the major challenge of today’s research. We present an improved patch-based fast noise level estimation using DCT and standard deviation method for fast and reliable noise level estimation and the result is compared with the available state-of-art methods. Experimental result shows the proposed method provides greater accuracy, the stability and also the proposed method is an average of six times faster than that of the state- of – art methods for noise level estimation.

Low-dose computed tomography image denoising based on joint wavelet and sparse representation.

One of the most challenging issues in low dose computed tomography (CT) imaging is image denoising and signal enhancement. Sparse representational methods have shown initial promise for these applications. In this thesis we present a wavelet based sparse representation denoising technique utilizing dictionary learning and clustering. By using wavelets we extract the most suitable features in the images to obtain accurate dictionary atoms for the denoising algorithm. To achieve improved results we also lower the number of clusters which reduces computational complexity. In addition, a single image noise level estimation is developed to update the cluster centers in higher PSNRs. A new image enhancement technique is developed for low-dose CT images to improve the quality of image for diagnostic purpose and reduce the blurring artifacts. The accuracy along with the computational efficiency of the proposed algorithm are then compared with recent approaches and clearly demonstrate the improvement of the proposed algorithm proposed in this thesis.

Low-dose computed tomography image denoising based on joint wavelet and sparse representation.

One of the most challenging issues in low dose computed tomography (CT) imaging is image denoising and signal enhancement. Sparse representational methods have shown initial promise for these applications. In this thesis we present a wavelet based sparse representation denoising technique utilizing dictionary learning and clustering. By using wavelets we extract the most suitable features in the images to obtain accurate dictionary atoms for the denoising algorithm. To achieve improved results we also lower the number of clusters which reduces computational complexity. In addition, a single image noise level estimation is developed to update the cluster centers in higher PSNRs. A new image enhancement technique is developed for low-dose CT images to improve the quality of image for diagnostic purpose and reduce the blurring artifacts. The accuracy along with the computational efficiency of the proposed algorithm are then compared with recent approaches and clearly demonstrate the improvement of the proposed algorithm proposed in this thesis.