Approximation of DCT without multiplication in JPEG

An original algorithm for the approximation of the discrete cosine transform (DCT) without multiplication is presented. The method is based on a recursive generator of discrete harmonic signals. The "speed/accuracy" ratio can be tuned due to input specifications. The actual accuracy and error analysis depend on the length of the DCT. We analyse the possibility of using our algorithm in JPEG image compression. However, application of the approximation method is possible in all problems using DCT. Our new approach to the DCT approximation is applicable to other transforms such as the DFT.