Abstract:It has been shown that equivariant convolution is very helpful for many types of computer vision tasks. Recently, the 2D filter parametrization technique has played an important role for designing equivariant convolutions, and has achieved success in making use of rotation symmetry of images. However, the current filter parametrization strategy still has its evident drawbacks, where the most critical one lies in the accuracy problem of filter representation. To address this issue, in this paper we explore an ameliorated Fourier series expansion for 2D filters, and propose a new filter parametrization method based on it. The proposed filter parametrization method not only finely represents 2D filters with zero error when the filter is not rotated (similar as the classical Fourier series expansion), but also substantially alleviates the aliasing-effect-caused quality degradation when the filter is rotated (which usually arises in classical Fourier series expansion method). Accordingly, we construct a new equivariant convolution method based on the proposed filter parametrization method, named F-Conv. We prove that the equivariance of the proposed F-Conv is exact in the continuous domain, which becomes approximate only after discretization. Moreover, we provide theoretical error analysis for the case when the equivariance is approximate, showing that the approximation error is related to the mesh size and filter size. Extensive experiments show the superiority of the proposed method. Particularly, we adopt rotation equivariant convolution methods to a typical low-level image processing task, image super-resolution. It can be substantiated that the proposed F-Conv based method evidently outperforms classical convolution based methods. Compared with pervious filter parametrization based methods, the F-Conv performs more accurately on this low-level image processing task, reflecting its intrinsic capability of faithfully preserving rotation symmetries in local image features.

Convolution theorems for the free metaplectic transformation and its application

Short-time free metaplectic transform: Its relation to short-time Fourier transform in $ L^2(\mathbb R^n) $ and uncertainty principles

Quaternionic free metaplectic transformation

A Generalized Convolution Theorem for the Special Affine Fourier Transform and Its Application to Filtering

A Convolution and Product Theorem for the Linear Canonical Transform

Uncertainty Principles for the Short-time Free Metaplectic Transform

New Advanced Template-matching Algorithm

A Convolution and Correlation Theorem for the Linear Canonical Transform and Its Application

New Convolution Theorem for the Linear Canonical Transform and Its Translation Invariance Property

Generalized Convolution and Product Theorems Associated with Linear Canonical Transform.

Uncertainty principles for short‐time free metaplectic transformation

Convolution theorems associated with quaternion linear canonical transform and applications

Generalized convolution theorem associated with fractional Fourier transform

Fourier Series Expansion Based Filter Parametrization for Equivariant Convolutions

Generalized Convolution Properties Based on the Modified Mittag-Leffler Function

Properties and Convergence Analysis of Fm(M)Let Transform

Convolution Type of Metaplectic Cohen's Distribution Time-Frequency Analysis Theory, Method and Technology

CONVOLUTION THEOREM OF THE COMPLEX FRACTIONAL FOURIER TRANSFORMATION DERIVED BY TRANSFORMATION OF ENTANGLED STATES

Slice Fourier transform and convolutions

Finite free convolutions via Weingarten calculus

A Convolution-Based Fractional Transform