MPEG-4 Codec Performance using a Fast Integer IDCT

Several standardized video coding algorithms use a well known discrete cosine transform (DCT) at the encoder to remove redundancy from video random processes. Its inverse, the IDCT is present at the decoder as well as at the encoder loop. The accuracy of this inverse transform, required to avoid large drift between the encoder and decoder, is defined in an IEEE standard which will be withdrawn from MPEG and ITU standardized codecs. Due to the huge number of computations required to compute the FDCT/IDCT (forward/inverse DCT) pair, reduction of their complexity is essential to speed up the video processing. In this paper, we propose a fast integer IDCT calculation method. Additionally, we insert it into the MPEG-4 (Part 2) reference software in order to validate our proposed method. The testing results using two different video sequences, at QCIF resolution, show similar PSNR average values between the reference MPEG-4 and the proposed MPEG-4 codec