Abstract:Typical non-local edge-aware filtering methods build long-range connections by deriving a minimum spanning tree (MST) from input image, and the structure of MST completely depends on the intensity difference between neighboring pixels. Each pixel on the MST only connects to a sub-set of pixels in 8-neighborhood, resulting in unbalanced information propagation along eight directions. Thus they tend to generate piece-wise constant filtering output with fake edges among sub-trees. In this paper, we propose two complementary trees to aggregate information from entire image. The structure of each tree depends on the spatial relationships of neighboring pixels. The distances between any two pixels in both spatial space and intensity space are the shortest distance on each tree. Novel algorithms are presented to recursively compute output and normalization constant on each tree in linear time. For each pixel, we first compute the outputs along eight sub-trees, namely sub-trees in left, top, right, bottom, top-left, top-right, bottom-left and bottom-right of the root node, and then fuse the outputs to obtain filtering output on each tree. The final filtering output of our method is simply averaging the outputs of two complementary trees. Moreover, we present a distance mapping scheme in range space to map the intensity difference between neighboring pixels to a new value, which enable our method to manageably filter out low-amplitude structures while sharpening major edges. Extensive experiments in graphic applications, such as image denoising, JPEG artifacts removal, and colorization, to demonstrate the effectiveness and versatility of our method.

Joint Contour Filtering

Path-Based Analysis for Structure-Preserving Image Filtering

Anisotropic Joint Trilateral Rolling Filter for Image Smoothing

Contour feature extraction from fuzzy image

Path-Based Analysis for Structure-Preserving Ima G E Filtering

Image edge detection based on multi-scale fusion technology

Joint Image Filtering with Deep Convolutional Networks

Contour Detection Improved by Frequency Domain Filtering of Gradient Image

Edge-preserving Image Decomposition Via Joint Weighted Least Squares

Two-level joint local laplacian texture filtering

Enhanced Edge-Perceptual Guided Image Filtering

Static/Dynamic Filtering for Mesh Geometry

Scale Adaptive Structure Tensor Based Rolling Trilateral Filter

Learning to detect contours in natural images via biologically motivated schemes

Image Filtering Based On Mathematical Morphology And Visual Perception Principle

Edge-preserving Smoothing for Image Decomposition Via a Hybrid Approach

Edge-aware texture filtering with superpixels constraint

The trinary image tree: A flexible edge-aware multiscale structure

Multilevel Filtering Image Denoising Algorithm Based on Edge Information Fusion

Edge detection based on joint iteration ghost imaging

Linear Time Edge-Aware Filtering on Complementary Tree Structures