Long range diffusion with control of the directional differences

2021 ◽  
Vol 2021 (29) ◽  
pp. 323-327
Author(s):  
Ali Alsam ◽  
Hans Jakob Rivertz
Keyword(s):  
Long Range ◽  
Adaptive Filter ◽  
Spatially Adaptive

A fast, spatially adaptive filter for smoothing colour images while preserving edges is proposed. To preserve the edges, we use a constraint that prohibits the increasing of the gradients in the process of diffusion. This constraint is shown to be very effective in preserving details and flexible in cases where more smoothing is desired. In addition, a filter of exponentially increasing diameter is used to allow averaging non-adjacent pixels, including those separated by strong edges.

scholarly journals Spatially-Adaptive Filter Units for Compact and Efficient Deep Neural Networks

2020 ◽  
Vol 128 (8-9) ◽  
pp. 2049-2067 ◽  
Author(s):  
Domen Tabernik ◽  
Matej Kristan ◽  
Aleš Leonardis
Keyword(s):  
Neural Networks ◽  
Adaptive Filter ◽  
Deep Neural Networks ◽  
Spatially Adaptive

scholarly journals A Sketch-Transformer Network for Face Photo-Sketch Synthesis

2021 ◽  
Author(s):  
Mingrui Zhu ◽  
Changcheng Liang ◽  
Nannan Wang ◽  
Xiaoyu Wang ◽  
Zhifeng Li ◽  
...  
Keyword(s):  
Long Range ◽  
State Of The Art ◽  
Structural Information ◽  
Generative Adversarial Networks ◽  
Spatial Dependency ◽  
Facial Image ◽  
Multi Scale ◽  
Adversarial Networks ◽  
Spatially Adaptive ◽  
Sketch Synthesis

We present a face photo-sketch synthesis model, which converts a face photo into an artistic face sketch or recover a photo-realistic facial image from a sketch portrait. Recent progress has been made by convolutional neural networks (CNNs) and generative adversarial networks (GANs), so that promising results can be obtained through real-time end-to-end architectures. However, convolutional architectures tend to focus on local information and neglect long-range spatial dependency, which limits the ability of existing approaches in keeping global structural information. In this paper, we propose a Sketch-Transformer network for face photo-sketch synthesis, which consists of three closely-related modules, including a multi-scale feature and position encoder for patch-level feature and position embedding, a self-attention module for capturing long-range spatial dependency, and a multi-scale spatially-adaptive de-normalization decoder for image reconstruction. Such a design enables the model to generate reasonable detail texture while maintaining global structural information. Extensive experiments show that the proposed method achieves significant improvements over state-of-the-art approaches on both quantitative and qualitative evaluations.

Automated 3-D cell population analysis in thick tissue sections using laser-scanning confocal-microscopy data

1993 ◽  
Vol 51 ◽  
pp. 562-563
Author(s):  
Hakan Ancin
Keyword(s):  
Laser Scanning ◽  
Population Analysis ◽  
Three Dimensional ◽  
Large Data ◽  
Laser Scanning Confocal Microscopy ◽  
Tissue Slices ◽  
Scanning Confocal Microscopy ◽  
Tissue Sections ◽  
Spatially Adaptive ◽  
Microscopy Data

This paper presents methods for performing detailed quantitative automated three dimensional (3-D) analysis of cell populations in thick tissue sections while preserving the relative 3-D locations of cells. Specifically, the method disambiguates overlapping clusters of cells, and accurately measures the volume, 3-D location, and shape parameters for each cell. Finally, the entire population of cells is analyzed to detect patterns and groupings with respect to various combinations of cell properties. All of the above is accomplished with zero subjective bias.In this method, a laser-scanning confocal light microscope (LSCM) is used to collect optical sections through the entire thickness (100 - 500μm) of fluorescently-labelled tissue slices. The acquired stack of optical slices is first subjected to axial deblurring using the expectation maximization (EM) algorithm. The resulting isotropic 3-D image is segmented using a spatially-adaptive Poisson based image segmentation algorithm with region-dependent smoothing parameters. Extracting the voxels that were labelled as "foreground" into an active voxel data structure results in a large data reduction.

Sign in / Sign up

Export Citation Format

Share Document