New dyadic transform for image compression

In this paper, a new order-8 dyadic transform (DT) is proposed for image compression. The proposed DT can be computed using a fast structural algorithm with operations of addition and binary shift only. Performance evaluations using a statistical model and real images show that the proposed DT performs better than the two DTs proposed by Lo and Cham and very close to DCT. Simulation results also show that there is a certain degree of compatibility between DCT and the proposed DT. In other words, DCT can be replaced by the proposed DT in existing image coding systems with little degradation in image quality.