Evaluation and Performance Comparison between JPEG2000 and SVC

Abstract

The scalable extension to the H.264/AVC has been established as a scalable video coding (SVC) standard, and JPEG2000 is a state of the act still image coding standard. The most distinctive difference between SVC and JPEG2000 is, in terms of spatial-frequency transformation, that SVC is based on the Discrete Cosine Transform (DCT) and JPEG2000 is based on Discrete Wavelet Transform (DWT). JPEG2000 outperforms its predecessor-JPEG which adopts DCT by nearly 30% in terms of compression performance. Despite the DWT has been successfully employed in still image coding, the latest video coding standard keeps on ultilising DCT and the relevant techniques as its coding tools. In this paper, we investigate SVC for intra frame coding and JPEG2000 for still image coding, on video sequences with different resolutions and features. From the experimental evaluations, we attempt to gain deeper insight into the application scenarios of these two transform techniques. The experimental results demonstrate that at high bitrates, JPEG2000 outperforms SVC, and it is vice versa. We also note that the PSNR difference between SVC and JPEG2000 decreases with increasing the picture resolution. Therefore, JPEG2000 is more suited for high definition picture compression. The spatial scalability of SVC and JPEG2000 is also investigated in the evaluation, and the results show that SVC gains quite better performance than JPEG2000.