China
Africa
Explore chapters and articles related to this topic
Design Considerations for Photon-Counting Detectors
Published in Katsuyuki Taguchi, Ira Blevis, Krzysztof Iniewski, Spectral, Photon Counting Computed Tomography, 2020
Figure 17.7 shows a simple example. The bilateral filter, a simple edge-preserving smoothing technique, is used as a form of nonlinear reconstruction (58). The bilateral filter is a variation of traditional Gaussian smoothing which adapts its smoothing kernel based on both spatial proximity and intensity proximity. Traditional Gaussian smoothing uses only spatial proximity, but by including intensity proximity (similarity of the CT numbers of the two pixels), the bilateral filter avoids smoothing together hard edges. Other, more sophisticated iterative reconstruction packages use optimization of a cost function, which is computationally intensive. While it is impossible to speak broadly of all iterative reconstruction packages, Figure 17.7 illustrates that nonlinear reconstruction methods should be tested carefully. For non-spectral iterative reconstruction, a wide range of tools, including model observers (13) and clinical studies (59), have replaced Fourier metrics such as NPS and MTF. Similar care may be needed to fully characterize spectral nonlinear reconstruction methods.
Digital Image Processing for Machine Vision Applications
Published in Sheila Anand, L. Priya, A Guide for Machine Vision in Quality Control, 2019
The bilateral filter is also defined as a weighted average of nearby pixels, in a manner very similar to Gaussian convolution. The difference is that the bilateral filter takes into account the difference in value with the neighbors to preserve edges while smoothing. The key idea of the bilateral filter is that for a pixel to influence another pixel, it should not only occupy a nearby location but also have a similar value. In order words, the rationale of bilateral filtering is that two pixels are close to each other not only if they occupy nearby spatial locations but also if they have some similarity in the photometric range, that is, pixel intensity. The bilateral filter, denoted by BF[I]P, is defined by the following equation: BF[I]p=1Wp∑q∈sGσs(‖p−q‖)Gσr(Ip−Iq)Iq
Surface Features
Published in Wolfgang Osten, Optical Inspection of Microsystems, 2019
where both f and g are normally Gaussian distribution functions. The bilateral filter has better edge-preserving performance than a Gaussian filter. However, the bilateral filter has some limitations, such as introducing staircase effect and some false edges.
Optimized Haar wavelet-based blood cell image denoising with improved multiverse optimization algorithm
Published in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2023
M. Mohana Dhas, N. Suresh Singh
Trilateral filter (Zhang et al. 2016) diminishes the noise in dark area of the image with lesser constraints defined by the user. Trilateral filter is another version of bilateral filter that is designed by enhancing the bilateral filter (Dai et al. 2017; Joseph and Periyasamy 2018). The range filter, impulse weight function and domain filter combine together to form a nonlinear spatial domain filter known as trilateral filter (Mansoor et al. 2014). For eradicating the blend of impulse noise and Gaussian noise, the trilateral filter is created. The trilateral filter preserves the edge details of the blood cell image and it also helps to maintain the quality of the output image. The denoised image obtained from the proposed IMVO – AT is then passed through the trilateral filter to smooth a low-SNR and thus enhances the performance of the denoised image.
Chronological-hybrid optimization enabled deep learning for boundary segmentation and osteoporosis classification using femur bone
Published in The Imaging Science Journal, 2023
Kiran Dhanaji Kale, Bharati Ainapure, Sowjanya Nagulapati, Lata Sankpal, Babasaheb Sambhajirao Satpute
The image acquired from the database is forwarded to pre-processing for making the image appropriate for further processing. Here, pre-processing is carried out using the Bilateral filter [35] to discard the unwanted noise in the input image. A bilateral filter is a modest and non-iterative technique, which combines the neighbouring non-linear value in the image to smoothen the noisy image and preserve the edges. It combines the grey level value of the neighbouring values based on geometric proximity and photometric similarities. Similar to the Gaussian convolution, the bilateral filter determines the mean of the pixels along with the difference in intensities, thereby preserving the edge information. The bilateral filter can be given by the given expression,
Resnet Features and Optimization Enabled Deep Learning for Indoor Object Detection and Object Recognition
Published in Cybernetics and Systems, 2022
Pre-processing is the process of removing unwanted calamities that exist in the image. In this research, the input image is used for pre-processing. Here, the pre-processing is done with a bilateral filter (Paris et al. 2009), which is a kind of non-linear filter. The advantage of a bilateral filter is that it preserves the edges of an image, reduces the noise and it smoothens the images, thereby it improves the quality of images. In addition, the bilateral filter exchanges the intensity of each pixel with a weighted average of intensity values from neighboring pixels. The expression for Bilateral Filtering is denoted as, where denotes the normalization factor, which confirms the totality of pixel weights as 1, and is represented as, where and indicates the filtering quantity for the image specifies the spatial Gaussian weighting, denotes the Gaussian range and indicates the pixel values. Thus, the pre-processed image is notated as