Scale and Statistics in Variable Conductance Diffusion
نویسنده
چکیده
Non-linear diffusion or Variable Conductance Diffusion (VCD) processes can be used as to filter away noise while preserving boundary information. However, adjustment of the control parameters is not well understood. By approaching the control of VCD systems through a multiscale statistical analysis, information about the image is revealed regarding the strength of boundaries and their influence with regard to the local geometry of the object. This information can then be used to monitor and guide the diffusion process.
منابع مشابه
A Numerical Modeling of a Variable Conductance Thermosyphon by Use of a Binary Mixture
In this paper the steady state performance of a variable conductance heat pipe, which its working fluid consists of two components (R11+ R113) is modeled. The role of the liquid film momentum and axial normal stresses are considered by use of the mean property of mixture and the governing equations are solved by conventional numerical methods. The results of the present model have been compared...
متن کاملUsing Statistical Pattern Recognition Techniques to Control Variable Conductance Diffusion
We present an approach for controlling relaxation parameters in variable conductance diffusion. This approach incorporates a Bayesian classifier to perform a partial labeling of an image, followed by a diffusion step. Conductance values between pixels are controlled by statistical measurements made of the partial classification. Several iterations follow, interleaving partial labeling with diff...
متن کاملKAZE Features
In this paper, we introduce KAZE features, a novel multiscale 2D feature detection and description algorithm in nonlinear scale spaces. Previous approaches detect and describe features at different scale levels by building or approximating the Gaussian scale space of an image. However, Gaussian blurring does not respect the natural boundaries of objects and smoothes to the same degree both deta...
متن کاملSpectra - limited Diffusion in the Characterization of Textural Patches
Ross T. Whitaker Department of Computer Science, University of North Carolina Chapel Hill, North Carolina, 27514 Image segmentation on the basis of texture is an important aspect of natural vision as well as a widely studied problem in artificial vision. This paper presents a method for discriminating patches in digital images on the basis of homogeniety of local properties within the image. Th...
متن کاملAnomalous conductance and diffusion in complex networks
We study transport properties such as conductance and diffusion of complex networks such as scale-free and Erdős-Rényi networks. We consider the equivalent conductance G between two arbitrarily chosen nodes of random scale-free networks with degree distribution P (k) ∼ k−λ and Erdős-Rényi networks in which each link has the same unit resistance. Our theoretical analysis for scale-free networks ...
متن کامل