Efficient multiplier-less parametric integer approximate transform based on 16-points DCT for image compression

Nabila Brahimi, Toufik Bouden, Tahar Brahimi, Larbi Boubchir

Abstract: The Discrete Cosine Transform (DCT) is one of the major components in most image and video compression systems. In this paper, a novel approach for developing integer approximation of the two-dimension 8 points conventional DCT-II is introduced. The proposed method is based on 16 points DCT-II and rounding off operations. As result, we obtained multiplication-free 8-points approximation of conventional DCT-II, noted $${\mathrm{T}}_{\mathrm{pa}}$$ T pa , with the parameter a = {1,2}, and its basic elements are {0, ± 1, ± 2}; no multiplication operations are required. These two transforms are the more efficient DCT approximations with a reduction in the term of complexity computation, requiring 24 additions and 2-bit shifts for a = 1 or 24 additions and 6-bit shifts for a = 2; no multiplication required. The comparative study carried-out to assess the performance of the proposed transform matrices against some popular existing ones using different quality measures, has shown that the proposed transforms outperform the state-of-the-art of DCT approximations at low and high bit rates, as well as providing a lower computational complexity. Therefore, the capability of proposed transforms matrices is improved in image compression application.