China
Africa
Explore chapters and articles related to this topic
Acquisition and Computation for Data in Biometric System
Published in Karm Veer Arya, Robin Singh Bhadoria, The Biometric Computing, 2019
Varadarajan et al. [47] proposed Chirp Z-Transform and Goertzel algorithms as preprocessing, block-based feature extraction and exponential binary particle swarm optimization for feature selection. For illumination normalization, combined approach of CZT and Goertzel algorithm is used. In Ref. [48], two techniques, anisotropic diffusion-based preprocessing and Gabor filter-based feature extraction, have been proposed. Anisotropic diffusion is used to preserve the edges for smoothing and enhancement. To capture facial features of specific angles Gabor filter is used. To search the feature space for the optimal feature subset, a binary particle swarm optimization-based feature selection algorithm is used. Thus improvement in efficiency is observed by using anisotropic diffusion. Varun et al. [49] developed a feature extraction method based on Hough Transform peaks. For efficient feature extraction, Block-wise Hough Transform peaks are used, and to search the feature space for the optimal feature subset, a Binary Particle Swarm Optimization (BPSO)-based feature selection algorithm is used.
Image Processing
Published in R. Suganya, S. Rajaram, A. Sheik Abdullah, Big Data in Medical Image Processing, 2018
R. Suganya, S. Rajaram, A. Sheik Abdullah
In anisotropic diffusion the main motto is to encourage smoothening within the region in preference to the smoothening across the edges. This is achieved by setting the conduction coefficient as 1 within the region and as 0 near edges, however the main problem involved in this is the detection of the presence and absence of edges. As a solution for this problem it is identified that conduction coefficient, if chosen locally as a function of magnitude of the gradient of the brightness function of the image, the edges can be determined. A general expression for anisotropic diffusion can be written as I(x, 0) = I0∂l∂t=div(F)+β(I0−I)
A
Published in Phillip A. Laplante, Dictionary of Computer Science, Engineering, and Technology, 2017
anisotropic diffusion a process of progressive image smoothing as a function of a time variable t, such that the degree and orientation of smoothing at a point vary according to certain parameters measured at that point (e.g., gray-level gradient, curvature, etc.) in order to smooth image noise while preserving crisp edges. The progressively smoothed image I(x, y, t)(where x, y are spatial coordinates and t is time) satisfies the differential equation ∂I/∂t=div(c∇I),
R2F-UGCGAN: a regional fusion factor-based union gradient and contrast generative adversarial network for infrared and visible image fusion
Published in Journal of Modern Optics, 2023
Yuning Xie, Gang Liu, Rui Xu, Durga Prasad Bavirisetti, Haojie Tang, Mengliang Xing
The anisotropic diffusion method has been successfully used in image multi-scale description, edge detection, image segmentation, and image enhancement [35]. It has achieved remarkable results in the field of image processing [36]. Perona and Malik et al. [37] were the first to put forward the anisotropic diffusion PM model, which uses the local features of the image to determine the diffusion coefficient. It can anisotropically model image noise distribution and remove them while keep saliance image feature such as edges, or contours, etc. A large number of studies show that anisotropic diffusion can not only filter the noise but also retain the edge detail information of the source image and even enhance the edge information of the image. Tsiotsios et al. [38] studied the parameters of anisotropic diffusion and proposed an automatic stop criterion. This criterion improves the smoothness of images and preserves the integrity of the edge. The equation of anisotropic diffusion can be expressed as Equation (1): where is the diffusion rate; t is time; It is an image that changes with time t.
Singular value decomposition-based anisotropic diffusion for fusion of infrared and visible images
Published in International Journal of Image and Data Fusion, 2019
Each source image is filtered using anisotropic diffusion process to extract base and detail layers. Useful information from base and detail layers is integrated into the fused image. The anisotropic diffusion process will smooth a given image at homogeneous regions whilst preserving the non-homogeneous regions (edges) using partial differential equations. It overcomes the drawbacks of isotropic diffusion. Isotropic diffusion uses inter-region smoothing. Thus, edge information is lost. In contrast, anisotropic diffusion uses intra-region smoothing to generate coarser resolution images. At each coarser, resolution edges are sharp and meaningful. Neither IR nor visible image provides the complete information about the scene. By the process of fusion, one can attempt to perceive complete information about the scene, especially for steamboat and man.
Modified CNN Architecture for Efficient Classification of Glioma Brain Tumour
Published in IETE Journal of Research, 2022
J. Angelin Jeba, S. Nirmala Devi, M. Meena
The input MRI images from the dataset are resized to (227 × 227). To preserve the fine details and decrease noise present in MRI images, literature papers have used filters such as median filter, diffusion filter and adaptive filter at the pre-processing stage. Among these Anisotropic diffusion filter has shown improved performance practically in terms of Peak Signal-to-Noise Ratio (PSNR) and Mean-Square-Error (MSE) values. Anisotropic diffusion filter assumes the Gaussian distribution for noise hence capable to improve the Signal to Noise (SNR) in the regions corrupted due to Gaussian noise. Anisotropic diffusion filter performs better by expelling noise and smoothening the image by saving the required edges and structures.