Catálogo de publicaciones - libros

In this paper we propose a novel feature extracting method based on spatiotemporal slice analyzing. To date, video features are focused on the character of every single video frame. With our method, the video content is no longer represented with every single frame. The temporal variation of visual information is taken as an important feature of video in our method. We examined this kind of feature with experiments in this paper. The experiment results show that the proposed feature is effective and robust for variant video content and format.

In this paper, we propose fast color interpolation methods for color filter array (CFA). The proposed methods exploit correlations among the three color components. It has been reported that the quality of color interpolation can be significantly improved by utilizing correlation among color components. However, the complexity of such methods may increase significantly compared to those methods which separately interpolate each color component. On the other hand, the major application of CFA is portable imaging devices such as digital cameras or mobile phones. For portable devices, power consumption is an important issue. In this paper, we propose high quality color interpolation methods with reduced complexity. Experimental results show that the proposed methods provide noticeable improvements without significantly increasing the system complexity.

Although palmprint based authentication approaches have shown promising results recently, efforts are still required to achieve higher performance with low time complexity for their use in real-life high security applications. Based on these requirements, the concept of two level matching scheme using line matching in Hough space and Curve segment Hausdorff Distance (CsHD) are explored and developed for speedy and accurate palmprint identification in this paper. This method combines the advantages of both local and global feature approaches, which is able to not only absorb the variance of palmprints by transforming the palmprint pattern into a new domain, but also overcome the problem of high computation complexity when the database size increases. The system employs low-resolution palmprint images captured by normal digital camera and achieves higher identification accuracy with lower time complexity.

In this paper, we mainly investigate how to improve the streaming quality of service over Peer-to-Peer (P2P) networks with the benefit of practical service traces. Considering that the QoS of P2P live streaming is generally affected by high dynamics in peer community, we abstract the online duration model of end users and then exploit this characteristic to improve the stability and encourage contribution in P2P networks, which in turn facilitates the quality of streaming.

The first effort involves the online information when constructing video delivery tree and hence achieves less playback disruption. The other algorithm takes the online duration into account when selecting neighbor peers for data exchange and advocates more mutual cooperation between peers. The experiment results show that the proposed schemes could respectively achieve higher stability and better delivery quality.

‘Coopetitive’ interaction strategy has been shown to give better results than similar strategies like ‘only cooperation’, ‘only competition’ etc [7]. However, this has been studied only in the context of visual sensors and for handling non-simultaneous events. In this paper, we study this ‘coopetitive’ strategy from a multimedia surveillance system perspective, wherein the system needs to utilize multiple heterogeneous sensors and also handle multiple simultaneous events. Applying such an interaction strategy to multimedia surveillance systems is challenging because heterogeneous sensors have different capabilities for performing different sub-tasks as well as dissimilar response times. We adopt a merit-cum-availability based approach to allocate various sub-tasks to the competing sensors which eventually cooperate to achieve the specified system goal. Also, a ‘coopetition’ based strategy is adopted for effectively utilizing the information coming asynchronously from different data sources. Multiple simultaneous events (e.g. multiple intrusions) are handled by adopting a predictive strategy which estimates the exit time for each intruder and then uses this information for enhanced scheduling. The results obtained for two sets of surveillance experiments conducted with two active cameras and a motion sensor grid are promising.

Pattern-based video coding representing moving regions in macroblock has very good potential for improved coding efficiency over existing standard H.264/AVC in the very low bit-rate range. However, the coding efficiency diminishes compared to H.264 with the bit rates because variable block sizes of H.264 for motion estimation and compensation are successfully selected for any bit rates using weighting between bit rate and image quality. The existing pattern-based video coding scheme could not handle this flexibility with bit rates due to its fixed size and lack of verification with other modes of H.264. In this paper we propose two large and one small pre-defined pattern sets and treat them as extra modes of H.264. We also investigate the weighting factor between bit rate and image quality compared to the existing one. The experimental results confirm the superiority of our approach compared to H.264 and other existing pattern-based video coding for not only the very low bit rate range but also wide range of bit rates.

Standard image and video techniques designed to provide the bitstream with some level of protection against channel errors usually add a controlled amount of redundancy. This redundancy may take the form of resynchronization makers which enable the decoder to restart the decoding process from a known state in the event of transmission errors. The Error Resilient Entropy Code(EREC) is a well known algorithm designed to reduce the added redundant information. Wavelet tree coding provides the most efficient of image compression currently available with respect to rate-distortion. This paper presents a performance enhancement of EREC on the bitstream of SPIHT, which greatly improves its ability to maintain the compressed image quality in the event of random errors. This paper demonstrates that Efficient and Robust EREC(EREREC) improves the image quality of SPIHT in the presence of channel errors. The simulation result shows that the quality of transmitted image is improved when compared to the existing EREC on SPIHT.

In this paper, we propose a deeply parallel architecture called Fragment-based Interleaving Dual Pipelines(FIDP) which can exploit all parallelisms in Lifting-based 2D DWT algorithm. FIDP adopts a fragment-based samples consumption policy and consists of two row processors and two column processors. These processors are organized as interleaving dual pipelines to operate effectively. FIDP takes N/4+N/2+1 cycles to finish a N×N 2D DWT while requires only 5N+2 buffer.

The convergence of response functions of memoryless window-based congestion control protocols is analyzed in this paper. All the control schemes (or response functions) is classified into 4 types according to the monotony of their response functions. Sufficient conditions are established for the convergence of Type 1 scheme, which includes most of the existing protocols such as additive-increase multiplicative-decrease (AIMD), generalized AIMD, binomial congestion control (BCC), Choose Your Response Function (CYRF) and so on. Type 4 scheme is not convergent while Type 2 and Type 3 schemes are conditional convergent, and their convergent conditions are derived. Furthermore, the proposed scheme can be made TCP-friendly and TCP-compatible. The proposed scheme can also be used to design multimedia friendly congestion control protocols. Thus the proposed result is very useful for extending the network service to different applications.

This paper deals with the problem of estimating the effort required to maintain a static pose by human beings. The problem is important in developing dance summarization and rehabilitation applications. We estimate the human pose effort using two kinds of body constraints – skeletal constraints and gravitational constraints. The extracted features are combined together using SVM regression to estimate the pose effort. We tested our algorithm on 55 dance poses with different annotated efforts with excellent results. Our user studies additionally validate our approach.