Two-step text detection framework in natural scenes based on Pseudo-Zernike moments and CNN

Guezouli, Larbi

doi:10.1007/s11042-022-13690-6

Cited by 4 publications

(1 citation statement)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moment kernels depend on the polynomial basis and might be represented in a continuous or discrete model. Legendre [14], Zernike [15,16], Gaussian-Hermite [17], pseudo-Zernike [18,19], and Fourier-Mellin [20,21] moments are some examples of continuous orthogonal moments. These continuous moments are not suitable for digital image analysis due to the digital nature of image pixel values and they need to be discretised along the image dimensions.…”

Section: Introductionmentioning

confidence: 99%

Four-Term Recurrence for Fast Krawtchouk Moments Using Clenshaw Algorithm

Asli

Rezaei²

2023

Electronics

View full text Add to dashboard Cite

Krawtchouk polynomials (KPs) are discrete orthogonal polynomials associated with the Gauss hypergeometric functions. These polynomials and their generated moments in 1D or 2D formats play an important role in information and coding theories, signal and image processing tools, image watermarking, and pattern recognition. In this paper, we introduce a new four-term recurrence relation to compute KPs compared to their ordinary recursions (three-term) and analyse the proposed algorithm speed. Moreover, we use Clenshaw’s technique to accelerate the computation procedure of the Krawtchouk moments (KMs) using a fast digital filter structure to generate a lattice network for KPs calculation. The proposed method confirms the stability of KPs computation for higher orders and their signal reconstruction capabilities as well. The results show that the KMs calculation using the proposed combined method based on a four-term recursion and Clenshaw’s technique is reliable and fast compared to the existing recursions and fast KMs algorithms.

show abstract