Optimally Rotation-Equivariant Directional Derivative Kernels
نویسندگان
چکیده
We describe a framework for the design of directional derivative kernels for two-dimensional discrete signals in which we optimize a measure of rotation-equivariance in the Fourier domain. The formulation is applicable to rst-order and higher-order derivatives. We design a set of compact, separable, linear-phase derivative kernels of di erent orders and demonstrate their accuracy.
منابع مشابه
Discrete spherical means of directional derivatives and their applications
The goal of this paper is two-fold. First we study theoretical properties of discrete spherical means of directional derivatives of a function. Then we focus on the two-dimensional case and use discrete circular means to derive rotation-equivariant discrete approximations of linear and nonlinear firstand second-order differential operators. Applications to nonlinear filtering of digital images ...
متن کاملLearning Equivariant Functions with Matrix Valued Kernels
This paper presents a new class of matrix valued kernels that are ideally suited to learn vector valued equivariant functions. Matrix valued kernels are a natural generalization of the common notion of a kernel. We set the theoretical foundations of so called equivariant matrix valued kernels. We work out several properties of equivariant kernels, we give an interpretation of their behavior and...
متن کاملLearning Equivariant Functions with Matrix Valued Kernels - Theory and Applications
This paper presents a new class of matrix valued kernels, which are ideally suited to learn vector valued equivariant functions. Matrix valued kernels are a natural generalization of the common notion of a kernel. We set the theoretical foundations of so called equivariant matrix valued kernels. We work out several properties of equivariant kernels, we give an interpretation of their behavior a...
متن کاملRotation Invariant Object Detection with Matrix-Valued Kernels
This article presents a rotation invariant object detection system. We propose two main contributions. On the one hand we fuse two different approaches, the ’Parts and Structure’-model and the generalized Hough Transform. Local image patches cast votes for the locations of the parts of the searched object. Then, an explicit model is fitted to the estimated density functions for the different pa...
متن کاملRotation Invariant Classification of 3D Surface Textures using Photometric Stereo and Surface Magnitude Spectra
Many image-rotation invariant texture classification approaches have been presented. However, image rotation is not necessarily the same as surface rotation. This paper proposes a novel scheme that is surface-rotation invariant. It uses magnitude spectra of the partial derivatives of the surface obtained using photometric stereo. Unfortunately the partial derivative operator is directional. It ...
متن کامل