An efficient fast integer DCT transform for images compression with 16 additions only

Abstract

A new fast matrix is introduced for the discrete cosine transform (DCT), an important transform that is of particular interest in image compression. For an 8-point DCT only 16 additions are required by appropriately introducing some zeros in the 8×8 Signed DCT (SDCT) matrix taking into account the orthogonality conditions. The second originality of this work is the development of a suitable quantization matrix suitable for our proposed transform matrix. This quantization matrix improves the peak signal to noise ratio (PSNR) by about 0.7 dB compared to the JPEG quantization matrix.