Zero-Order Reverse Filtering

@article{Tao2017ZeroOrderRF,
  title={Zero-Order Reverse Filtering},
  author={Xin Tao and Chao Zhou and Xiaoyong Shen and Jue Wang and Jiaya Jia},
  journal={2017 IEEE International Conference on Computer Vision (ICCV)},
  year={2017},
  pages={222-230}
}
  • Xin Tao, Chao Zhou, Jiaya Jia
  • Published 13 April 2017
  • Engineering
  • 2017 IEEE International Conference on Computer Vision (ICCV)
In this paper, we study an unconventional but practically meaningful reversibility problem of commonly used image filters. [] Key Method A very simple yet effective zero-order algorithm is proposed. It is able to practically reverse most filters with low computational cost. We present quite a few experiments in the paper and supplementary file to thoroughly verify its performance. This method can also be generalized to solve other inverse problems and enables new applications.
Iterative first-order reverse image filtering
TLDR
An iterative first-order reverse image filtering algorithm based on the Newton-Raphson method, which could effectively reverse many prevalent image filters and probe the behaviors of black-box image filtering systems.
Reverse image filtering using total derivative approximation and accelerated gradient descent
TLDR
Results demonstrate that the proposed algorithm outperforms the state-of-the-art in that it is at the same level of complexity as that of the fastest reverse filter, but it can reverse a larger number of filters, and its complexity is much smaller.
Polyblur: Removing Mild Blur by Polynomial Reblurring
TLDR
The proposed algorithm first estimates image blur and then compensates for it by combining multiple applications of the estimated blur in a principled way, leading to superior results than other highly complex and computationally demanding techniques.
Adaptive Quantile Sparse Image (AQuaSI) Prior for Inverse Imaging Problems
TLDR
The Adaptive Quantile Sparse Image (AQuaSI) prior is proposed, which is based on a quantile filter, can be used as a joint filter on guidance data, and be readily plugged into a wide range of numerical optimization algorithms.
Alternately Guided Depth Super-resolution Using Weighted Least Squares and Zero-order Reverse Filtering
TLDR
This paper proposes alternately guided depth super-resolution using weighted least squares (WLS) and zero-order reverse filtering and shows that the proposed method outperforms state-of-the-art methods in terms of quantitative and qualitative measurements.
Bregman inverse filter
An inverse problem in which the observed signal is produced by passing the original signal through a filter that is derived from a convex minimisation problem is dealt. The filter is available as a
Learning in-place residual homogeneity for single image detail enhancement
TLDR
An image detail enhancement algorithm based on in-place residual homogeneity (IP), which gets the detail layer directly and amplifies it and has many good properties, such as being fast, edge-aware, robust, and parameter-free.
Lightness-aware contrast enhancement for images with different illumination conditions
TLDR
A novel contrast enhancing model is proposed, which is a simple linear fusion process based on an original image and its initial enhancement, which performs well on simultaneously improving image contrast and keeping its naturalness.
Recurrently-Trained Super-Resolution
TLDR
The work of using an objective quality measure in image enhancement is the first one whose validity was verified by showing similar results to the actual user's subjective evaluation and makes existing SR algorithms more useful by clearly improving the effect of SR.
...
...

References

SHOWING 1-10 OF 39 REFERENCES
Image smoothing via L0 gradient minimization
TLDR
This work presents a new image editing method, particularly effective for sharpening major edges by increasing the steepness of transition while eliminating a manageable degree of low-amplitude structures in an optimization framework making use of L0 gradient minimization.
Rolling Guidance Filter
TLDR
A new framework to filter images with the complete control of detail smoothing under a scale measure is proposed, based on a rolling guidance implemented in an iterative manner that converges quickly and achieves realtime performance and produces artifact-free results.
Fast High‐Dimensional Filtering Using the Permutohedral Lattice
TLDR
This algorithm is the first implementation of a high‐dimensional Gaussian filter that is both linear in input size and polynomial in dimensionality, and achieves a consistently high accuracy relative to ground truth (> 45 dB).
Edge-preserving decompositions for multi-scale tone and detail manipulation
TLDR
This paper advocates the use of an alternative edge-preserving smoothing operator, based on the weighted least squares optimization framework, which is particularly well suited for progressive coarsening of images and for multi-scale detail extraction.
A Machine Learning Approach for Non-blind Image Deconvolution
TLDR
This work relies on a two-step procedure, but learns the second step on a large dataset of natural images, using a neural network, and shows that this approach outperforms the current state-of-the-art on aLarge dataset of artificially blurred images.
Cross-Field Joint Image Restoration via Scale Map
TLDR
A scale map is introduced as a competent representation to explicitly model derivative-level confidence and new functions and a numerical solver are proposed to effectively infer it following new structural observations.
Adaptive manifolds for real-time high-dimensional filtering
TLDR
This work presents a technique for performing high-dimensional filtering of images and videos in real time by computing the filter's response at a reduced set of sampling points, and using these for interpolation at all N input pixels, and shows that for a proper choice of these sampled points, the total cost of the filtering operation is linear.
Structure-preserving image smoothing via region covariances
TLDR
This study proposes an alternative yet simple image smoothing approach which depends on covariance matrices of simple image features, aka the region covariances, and uses second order statistics as a patch descriptor to implicitly capture local structure and texture information.
100+ Times Faster Weighted Median Filter (WMF)
TLDR
A new joint-histogram representation, median tracking, and a new data structure that enables fast data access are proposed that reduce computation complexity from O(r2) to O( r) where r is the kernel size.
Image deblurring and denoising using color priors
TLDR
This work presents a novel image deconvolution algorithm that deblurs and denoises an image given a known shift-invariant blur kernel in a unified framework that can be used for deblurring, denoising, and upsampling.
...
...