FAST 8×8×8 RCF 3D_DCT/IDCT TRANSFORM FOR REAL TIME VIDEO COMPRESSION AND ITS FPGA IMPLEMENTATION

• Published in: International journal of advances in engineering and technology Vol. 7; no. 2; p. 292
• Main Authors: Hatim, Anas, Belkouch, Said, Hassani, Moha M'rabet
• Format: Journal Article
• Language: English
• Published: Bareilly I A E T Publishing Company 01.05.2014
• ISSN: 2231-1963

Video compression takes place increasingly in many applications which are constantly involving. It becomes more demanding in terms of performance at the expense of more power consumption. Discrete Cosine transform is the most common technique used in the compression field. In this paper we present an efficient fast Row Column Frame (RCF) 3D_DCT algorithm recently introduced in video compression. The optimization consists on the elimination of all the multiplications needed to compute the coefficients of the 3D DCT. The multiplications are gathered at the end of the 3D DCT and merged in the quantization cube. The mathematical demonstration and complexity comparisons with other techniques are presented, showing that new (RCF) 3D_DCT introduced makes important savings on arithmetic operations, even if there is a little decrease on the quality of the videos but it respects the video standards. We made 59% savings on the number of additions and eliminate totally (100%) the multiplications in reference to the standard RCF technique. The resulting transform is regular and symmetric. We present hardware architectures to implement the 3D DCT and detailed their properties. We proposed ping pong buffers as transpose memory to increase the performance of the architectures. An implementation on a 180 nm FPGA chip has been done and discussed at the end. We can process high resolution video standards like HDTV using our architectures with both reduced power and area.