No abstract

No abstract

No abstract

video coding scheme based on discrete wavelet transform (DWT), as it provides a better way to address scalability functionalities, than MPEG-2. 'Ib code wavelet coefficients efficiently, we use set partitioning in hierarchical trees (SPIHT) and adaptive arithmetic coding algorithms. Motion compensation (MC) is done in spatial domain to remove temporal redundancy present between frames. To avoid blocking artifacts caused by block motion compensation, overlapping block motion compensation (OBMC) is done. The proposed video encoder is compared with MPEG-2. I. I N T R O D U C T I O NThe discrete wavelet transform (DWT) has recently received a lot of attention due to the multiresolution representation it provides of a signal with localization in both time and frequency. This property is very desirable in image and video coding applications. The use of DWT to image coding application has become much more interesting after Shapiro introduced and popularized zerotree structure in [l] to code images at very high compression ratios with good quality. The embedded zerotree of wavelet coefficients (EZW) algorithm of [l] though gives better results, the set partitioning in hierarchical trees (SPIHT) [5] is recognized as one of the best methods available in literature to encode still images at target bit-rate/ distortion. SPIHT gives around 2 dB improvement over the EZW for same bit rate.In this paper we propose a new video coding technique based on DWT which gives quality of compressed video comparable to . A similar method based on zerotree concept has been proposed for very low bit rate video coding in [2]. The specific building blocks of the proposed video coder are: 1) DWT to remove spatial correlation present in each frame before and after motion compensation is done, 2) block motion compensation in spatial domain to reduce temporal redundancy present between frames, 3) overlapped blocked motion compensation to avoid blocking artifacts caused by block motion compensation so that cohesive input is presented for DWT calculations, 4) SPIHT in DWT domain to quantize wavelet coefficients and 5) adaptive arithmetic coder to code output given by SPIHT. Motion compensation is done in spatial domain on block-size 16x16 with half-pel accuracy. To avoid discontinuities at boundaries of each block arising from block motion compensation and which gives rise to high frequencies in wavelet transform domain, the overlapped block motion compensation (OBMC) of H.263 [3] is used so that cohesive input is presented to DWT calculations. The proposed method provides a better way to address the scalability functionalities than MPEG-2, because of multiresolution structure inherent in DWT. The proposed algorithm is compared with MPEG-2' for ballet video sequence of size 704x480, at 30 frames per second rate.Organization of the paper is as follows. In Section I1 a brief introduction of the SPIHT algorithm is given. Section 111 outlines the video coding and decoding algorithm. Simulation results axe given in Section IV, while Section V conclude...

Abstract