Abstract :
In this paper, we present fast efficient unified algorithms for the computation of one-dimensional (1-D) integer discrete cosine transform (DCT) and its inverse transform (IDCT), which can be used to compute different points (e.g. 4, 8, 16 and 32 point) transforms for the emerging video coding standard - High Efficiency Video Coding (HEVC). Since HEVC supports different transform unit (TU) sizes for the computation of DCT, ranging from 4-point to 32-point, it is necessary to develop an efficient reusable architecture which not only support the different transform size but also has low area cost of hardware implementation. The efficient 1-D DCT and 1-D IDCT algorithms are derived by decomposing the coefficient matrix of N-point transform into small size coefficient matrices based on the symmetric property of the integer transform matrices. The proposed algorithms are verified in Matlab and modeled in Verilog as well as implemented. With lower hardware complexity and better quality, the proposed algorithms can be applied to the emerging HEVC standard.
