The fast computation of DCT in JPEG algorithm

In this paper two new parallel algorithms are proposed for the computation of discrete cosine transform (DCT) in JPEG algorithm on 32-bit computers. These algorithms are twice as fast as the fastest known DCT algorithm published to date. This is possible because 16-bit wordlength is sufficient for the computation of DCT, so the arithmetic unit is used to perform simultaneously two 16-bit operations. All operations in the DCT algorithm, that are not suitable for parallel processing, are substituted by equivalent operations.