正交旋转不变 V 矩及其在图像重建中的应用

定义在单位圆盘上的正交旋转不变矩函数(如Zernike矩) 具有非常广泛的应用. 本文基于一类正交分段多项式函数系--V系统, 构造了一种新型的矩函数, 称之为正交旋转不变V矩(简称为V矩). 除了正交性、旋转不变性之外, 由于V系统具有次数低、表达式简单的优点, V矩能够避免传统矩函数中高阶多项式的计算, 从而能够保证数值稳定性, 降低计算复杂度. 实验结果表明, V矩比传统的正交旋转不变矩具有更好的图像重建与图像检索结果.

Orthogonal Rotation-invariant V Moments and Application to Image Reconstruction

Orthogonal rotation-invariant moments (ORIMs) such as Zernike moments introduced and defined on a continuous unit disk have been proven to be powerful tools in many applications. A new kind of ORIMs named V moments (VMs) is introduced in this paper, which is based on a class of orthogonal piecewise polynomial functions--V system. In addition to orthogonality and rotation-invariant, VMs have two special advantages of numerical stability and low computational complexity, which originate from the low degrees and simple expressions of the V system. Experimental results are presented to show that VMs can have a better image reconstruction and image retrieval results than the traditional moments.