Quaternion fractional-order color orthogonal moment-based image representation and recognition

Abstract Inspired by quaternion algebra and the idea of fractional-order transformation, we propose a new set generalized Laguerre orthogonal moments (QFr-GLMs) based on polynomials. Firstly, proposed QFr-GLMs are directly constructed in Cartesian coordinate space, avoiding need for conversion between polar coordinates; therefore, they better image descriptors than circularly coordinates. Moreover, unlike latest Zernike transformations, which extract only global features from color images, our can both local features. This paper also derives invariant color-image QFr-GLMs, enabling geometric-invariant pattern recognition images. Finally, performances moment invariants were evaluated simulation experiments correlated Both theoretical analysis experimental results demonstrate value their geometric representation