Fast splitting alpha-rooting method of image enhancement: tensor representation

Area of Science:

• Image Processing
• Signal Processing
• Computational Imaging

Background:

• Two-dimensional (2-D) images can be represented by one-dimensional (1-D) splitting-signals in tensor representation.
• Image enhancement often involves processing these spectral components, with methods like alpha-rooting traditionally requiring significant computation.

Purpose of the Study:

• To propose a fast implementation of the alpha-rooting image enhancement method.
• To leverage the tensor representation and discrete Fourier transform (DFT) for computational efficiency.

Main Methods:

• Utilizing the tensor representation of 2-D images, where splitting-signals carry spectral information.
• Implementing the alpha-rooting method by processing a single splitting-signal with respect to the discrete Fourier transform (DFT).
• Describing the fast algorithm in both frequency and spatial domains.

Main Results:

• The proposed method significantly reduces computational complexity compared to traditional alpha-rooting.
• The fast implementation uses two 1-D N-point DFTs instead of two 2-D N x N-point DFTs.
• Achieved a computationally efficient image enhancement process.

Conclusions:

• The tensor representation combined with DFT enables a faster alpha-rooting algorithm.
• This approach offers a more efficient way to perform spectral component modification for image enhancement.
• The proposed method provides a practical advancement in digital image processing techniques.