Error-resistance and Low-complexity Integer Inverse Discrete Cosine Transform

In this paper, a low-complexity multiplication-free integer inverse discrete cosine transform (IDCT) based on data flow structure of improved AAN’s fast IDCT is proposed for error drifting resistance of decoder. Two algorithms are used in this integer IDCT improvement. One is common factor extraction which extracts the complicated common factors from transform kernel to scale; the other is two-stage scale which splits a more than 16-bit scale into two less than 16-bit scales. With the two algorithms, high-accuracy integer IDCT is implemented in lower complexity. The experimental results show that the proposed transform exceeds the requirements of IEEE1180-1990 significantly (about 10 times). The results of the proposed IDCT implemented into MPEG-2 and MPEG-4 decoders instead of original 64-bit floating-point IDCT also show that it reduces the error drifting of decoders efficiently.