Image denoising using multiscale directional cosine bases

Motivated by the fact that in images, there is usually a presence of local strongly oriented harmonics, a representation which is both well-localised in frequency and orientation is desirable to efficiently describe such oriented harmonic features. Here we introduce a family of multiscale trigonometric bases for a bi-variate function called the multiscale directional cosine bases for image denoising tasks. Our results show the promise of the new bases which almost consistently outperform other image representation bases on natural images.