A low cost 2-D inverse discrete cosine transform design for image compression

This paper proposes a low cost 2-D inverse discrete cosine transform (IDCT) for image compression by using cyclic convolution and adder-based implementation. We formulate the 1-D IDCT into cyclic convolution and carry out the multiplications through adders by using the concept of common sub-expression sharing. The proposed 2-D IDCT design costs 7413 gates plus 1024 bits of memory with 66 MHz throughput based on a 0.35 /spl mu/m CMOS technology.