With the rapid improvements in digital communication technologies, distributing high-definition visual information has become more widespread. However, the available technologies were not sufficient to support the rising demand for high-definition video. This situation is further complicated when the network resources such as the available bandwidth fluctuates, or packet losses occur during transmission. In this dissertation we present several video compression techniques which are capable of adapting with the varying network conditions. We address both challenges namely, the fluctuations in the available resources such as the bandwidth and processing power, and packet losses. These problems in turn translates into degradation of the perceived video playback as jitter, and delay before video playback starts. Hence, we concentrate on developing robust and fast adaptive video coding schemes necessary for handling the changes in the physical characteristics of the communication networks. We present a new multi-layer scalable video coding (SVC) method for optimizing the bit-per-pixel rate of the video which is robust against packet losses. The method reduces the quality degradation in presence of data loss by re-organizing the frames in a hierarchical structure and improving the video quality through decomposing each frame suitably to restrict the error propagation. Moreover, we present a solution for the quality degradation in video reconstruction when the video is scrambled for privacy protection. We also present two methods based on multiple description video coding (MDC) to handle packet losses in networks with a high rate of transmission error. The proposed methods are based on combining SVC with MDC through decomposing the video into spatial sub-streams in the first method, and SNR sub-streams in the second method. In both proposed methods, the error resilience of the video is increased. The proposed methods have the capability of being used as SVC methods where any data loss or corruption reduces the quality of the video in a minimized way, and except for the case when all descriptions are lost, the video streams do not experience jitter at playback. The proposed methods provide the feasibility of reducing data rate by scaling down the video whenever the connection suffers from a low bandwidth problem. We also propose Discrete Wavelet Transform (DWT)-based optimizations for MDC. A major drawback in MDC methods is their inefficiency in terms of bit-per-pixel which is a consequence of preserving correlation between decomposed video segments. We propose a method based on the self-similarity between DWT coefficients at different frequency levels to improve the coding efficiency of DWT-based MDC. In the proposed method, whenever a description is lost the coefficients at the delivered descriptions are utilized for estimating the missing data using self-similarity property.@en

With video surveillance systems becoming ubiquitous nowadays, protecting people's privacy raises an increasingly serious concern. Video streaming with privacy protection requires modifying parts of the video content. This modification should provide the possibility of unprotected access to the video if the user is authenticated through a private key. However, any modification in the content of a video can result in a drift error and deteriorate the quality of the reconstructed video. In addition, it is required that the privacy protection does not adversely affect the computational complexity and the coding efficiency in terms of bitrate. In this work, we propose a drift-free method for scrambling the privacy protected regions of the frames while preserving the coding efficiency. Our proposed method provides the possibility of utilizing private keys for restricting unauthorized access to the private contents of the video with a small increase in the computational complexity of the encoder and decoder. The experimental results indicate that our proposed drift-free method can achieve a higher coding efficiency of 0.6 dB on average compared to similar methods.@en

With video surveillance systems becoming ubiquitous nowadays, protecting people's privacy raises an increasingly serious concern. Video streaming with privacy protection requires modifying parts of the video content. This modification should provide the possibility of unprotected access to the video if the user is authenticated through a private key. However, any modification in the content of a video can result in a drift error and deteriorate the quality of the reconstructed video. In addition, it is required that the privacy protection does not adversely affect the computational complexity and the coding efficiency in terms of bitrate. In this work, we propose a drift-free method for scrambling the privacy protected regions of the frames while preserving the coding efficiency. Our proposed method provides the possibility of utilizing private keys for restricting unauthorized access to the private contents of the video with a small increase in the computational complexity of the encoder and decoder. The experimental results indicate that our proposed drift-free method can achieve a higher coding efficiency of 0.6 dB on average compared to similar methods.@en

Video streaming over unreliable networks requires preventive measures to avoid quality deterioration in the presence of packet losses. However, these measures result in redundancy in the transmitted data which is utilized to estimate the missing packets lost in the delivered portions. In this paper, we have used the self-similarity property if the discrete wavelet transform (DWT) to minimize the redundancy and improve the fidelity of the delivered video streams in presence of data loss. Our proposed method decomposes the video into multiple descriptions after applying the DWT. The descriptions are organized in such a way that when one of them is lost during transmission, it is estimated using the delivered portions by means of self-similarity between the DWT coefficients. In our experiments, we compare video reconstruction in the presence of data loss in one or two descriptions. Based on the experimental results, we have ascertained that our estimation method for missing coefficients by means of self-similarity is able to improve the video quality by 2.14dB and 7.26dB in case of one description and two descriptions, respectively. Moreover, our proposed method outperforms the state-of-the-art Forward Error Correction (FEC) method in case of higher bit-rates.@en

Video streaming over unreliable networks requires preventive measures to avoid quality deterioration in the presence of packet losses. However, these measures result in redundancy in the transmitted data which is utilized to estimate the missing packets lost in the delivered portions. In this paper, we have used the self-similarity property if the discrete wavelet transform (DWT) to minimize the redundancy and improve the fidelity of the delivered video streams in presence of data loss. Our proposed method decomposes the video into multiple descriptions after applying the DWT. The descriptions are organized in such a way that when one of them is lost during transmission, it is estimated using the delivered portions by means of self-similarity between the DWT coefficients. In our experiments, we compare video reconstruction in the presence of data loss in one or two descriptions. Based on the experimental results, we have ascertained that our estimation method for missing coefficients by means of self-similarity is able to improve the video quality by 2.14dB and 7.26dB in case of one description and two descriptions, respectively. Moreover, our proposed method outperforms the state-of-the-art Forward Error Correction (FEC) method in case of higher bit-rates.@en