A Framework for H.264 with Video Enhancement Components for Minimizing Errors and Noise Mohamed T. Faheem Saidahmd Faculty of Engineering Taif Univ, KSA Amany Sarhan Faculty of Engineering, Tanta University, Egypt Rasha Orban Mahmoud Nile Institute of Commerce & Computer Technology, Mansoura , Egypt Abstract— Video transmission over error prone networks, such as the Internet and wireless networks, requires some methods of receiver-side video enhancement to achieve good quality. In this paper, we present a receiver-side video enhancement component to be augmented on the receiver (decoder) side to minimize errors and noise that exist in the received video. This component is characterized by using light-weight algorithms for enhancement in order to use less power which makes it suitable for battery-operated devices such as laptops and mobiles. Keywords-component; Spatial Video denoising, error concealment, video transmission, video compression I. INTRODUCTION The transmission of compressed video sequences over networks is potentially subjected to packet losses and noise existence [32]. In these applications, one has to cope with the fact that the network may be subject to congestion, thus causing some packets to be unusable at the receiver side because of excessive delay. Packet loss can be detrimental to real-time interactive video over lossy networks because one lost video packet can propagate errors to many subsequent video frames due to the encoding dependency between frames [4]. Two major problems may occur to the received video: noise and packet loss. Considerable effort is recently being spent in the development of effective techniques for both problems. Video denoising techniques work either spatially [3], temporally [14] or spatio-temporally [27]. However, temporal and spatio-temporal techniques require that the video sequence to be available ahead before starting the denoising process. This is not possible in applications that contain real-time video stream such as video conferencing or life broadcasting. Thus, the spatial techniques are more suitable for such applications as it process the frames as soon as they arrive. The most popular spatial video denoising techniques are: 2D Discrete Wavelet Transform (2D DWT) [5] and 2D Dual Tree Complex Wavelet Transform (2D DTCWT) [15]. In [6] we held an analytical study on both techniques from the point of view of the quality of the videos and the time required for denoising. From this analysis, we found out that DWT produces almost the same quality of the videos as DTCWT but in much less time especially in low to moderate levels of noise. As it is a challenge to obtain a good quality of the video in minimum time and thus less power and based on the analytical study and on the fact that low levels of noise is visually unnoticeable by human, we built an intelligent denoising system in [6] that achieves good quality videos in minimum time. The system first estimates the noise level in the frame then uses it to decide the proper action taken towards the frame. For the second source of error, many error recovery techniques have been proposed to repair damaged video due to packet loss. These techniques can be broadly categorized into three groups by whether the encoder or decoder plays the primary role, or both are involved in cooperation with each other [34]. Examples of error control techniques at the encoder side include Forward Error Correction (FEC) [21], joint source and channel coding (JSCC) [8], and layered coding [16]. The error controls that have interaction between encoder and decoder are called feedback based error control. Examples in this category include Retransmission [9], Reference Picture Selection (RPS) [11] and Intra Update [13]. Error control techniques at the decoder side include spatial and temporal smoothing [35], interpolation [19], and filtering [30]. In general, these techniques attempt to recover the damaged videos by estimation and interpolation [33]. However, decoder side error concealment is preferred as it does not require resending of packets which adds a burden on the network and also does not require encoder side evolvement which also relives the server of this type of workload. In our previous work [7], we proposed a set of decoder- side (i.e. receiver-side) spatial and temporal error concealment algorithms. We concentrated on building simple, yet efficient, algorithms to minimize the power consumed by requiring less computation time. From the results of comparison of these algorithms with each other and with a group of other error concealment algorithms in terms of quality of the videos and the time required for concealment, we concluded two of them (one for each type of concealment) to be the best algorithm as they give excellent tradeoff between high quality videos and consumed time for computation. With intention to remove both noise and errors from the received videos at high quality and in minimum time and minimum power consumed, in this paper we introduce a framework for H.264 decoder that contains enhancement components for removing noise and errors. These components uses low complexity – high output quality algorithms to achieve such goal. The main advantage of these component are their low complexity and minimum computation time which makes them suitable for battery operated devices. The rest of the paper is organized as follows: section II gives the basics of the spatial video denoising techniques and concentrates on the used techniques. Section III describes how the intelligent denoising component works. Section IV 978-1-4244-7042-6/10/$26.00 ©2010 IEEE 238