Study Of Image Compression Coding Based On Warped Discrete Cosine Transform

At present, the discrete cosine transform theory has been quite mature, and has been well applied in image and video coding. In image compression coding, DCT eliminate the spatial correlation of each pixel in image block, then the image has a good concentration of energy, lay the foundation for the compression. However, the DCT fails to transform much information into low-frequency components if the block of image contains mostly high-frequency components. With the frequency distribution of the input signal, Warped DCT makes the energy distribution is more suitable for image coding by warping the signal frequency, and has been widely applied in image compression and voice processing applications.This paper describes the theory of Warped DCT and the image-compression coding based on the Warped DCT. The Warped DCT is a cascade connection of a conventional DCT and first-order all-pass filters whose parameters can be adjusted to provide frequency warping. For each parameter, the corresponding Warped DCT matrices are computed a priori. For each image block, the best parameter is chosen from the set and the index is sent to the decoder as side information. At the decoder, Warped DCT inverse matrices are performed to reconstruct the image. The image compression coding based on Warped DCT outperforms the DCT-based compression coding, for high bit rate applications and for images with high-frequency components.The latter part of this paper introduces the improve algorithm of Warped DCT-based image compression coding. The algorithm of Warped DCT-based image compression considers distortion only, ignore the given bit budget. So improve the algorithm performance in the rate-distortion sense to improve the overall coding performance. Warped DCT improves the performance of conventional DCT by frequency warping, however, the Warped DCT has only been applicable to high bit-rate coding applications because of the overhead required to define the parameters of the warping filter. For low bit-rate applications, an algorithm that Warped DCT is used in conjunction with image down sampling-based block coding is employed.