China
Africa
Explore chapters and articles related to this topic
Iterative Reconstruction Methods in X-ray CT
Published in Paolo Russo, Handbook of X-ray Imaging, 2017
Geert Van Eyndhoven, Jan Sijbers
Various types of algorithms introduce a regularization term, U(x), in the objective function. The optimization problem then becomes x*=argminx(‖Wx−p‖2+λU(x)) where λ > 0 is the regularization parameter that controls the strength of the regularization. The regularization term, U(x), typically reflects some prior knowledge about the scanned object, x. Among many options, popular choices for U(x) include the total variation penalty (Chambolle 2004), smoothness priors (Tang et al. 2009), and the non-local means prior (Lou et al. 2010; Zhang et al. 2010).
Image speckle noise denoising by a multi-layer fusion enhancement method based on block matching and 3D filtering
Published in The Imaging Science Journal, 2019
Shuo Huang, Ping Zhou, Hao Shi, Yu Sun, Suiren Wan
In recent years, non-local algorithms for image denoising have been widely studied since they can obtain excellent image restoration effects [32–35]. One famous non-local algorithm is the non-local means (NL-means) method [32], which removes the zero mean additive gaussian white noise of a given pixel in an image by calculating the weighted average grey value of all pixels in the image. The weight of each pixel is determined by the similarity between the greyscale intensity vector of the pixels in a square window centred by it and that of pixels in the squire window with the same size centred by the pixel to be denoised. Since fully developed speckle noise can be regarded as zero-mean additive noise, these methods have good performance in removing fully developed speckle noise.
A marker-free method for structural dynamic displacement measurement based on optical flow
Published in Structure and Infrastructure Engineering, 2021
Jinsong Zhu, Ziyue Lu, Chi Zhang
Main image noise includes Gaussian noise and salt-and-pepper noise, and several methods were applied to image processing (Patidar, Gupta, Srivastava, & Nagawat, 2010). For instance, noise reduction can be carried out by mean filter, median filter or max/min filter. However, they destroy the edge information when removing noise. Particularly in displacement tracking, the accuracy depends more on a prominent gradient and edge information. As a denoising algorithm used in recent years, non-local means (N-L means) is to estimate the current pixel derived from a weighted average of the pixels in the image, according to the similarity between these pixels and the target pixel (Buades, Coll, & Morel, 2005).
A model for screening eye diseases using optical coherence tomography images
Published in International Journal of Computers and Applications, 2022
Sanchay Gupta, Siddharth Chandra, N. Maheswari, M. Sivagami
Non-Local Means Denoising is a method or procedure in image processing for denoising and filtering out various types of noises including both gaussian and salt-pepper noise in an image. Noise is especially categorized to be an arbitrary variable with zero mean. Considering a noisy pixel p = p0 + n, where is the true given value of the pixel and n is the given noise in that pixel. So we can take a large sum of the same pixels, say N, from a different set of images and calculates their average. Generally, we should get p = p0 because mean of collective noise is considered to be zero.