The aim of this work is to provide a fast approach for monocular SLAM initialization by constructing an initial 3-D map with interest points that are susceptible to be automatically tracked. Interest points' depth is inferred by means of a linear regression model, which estimates depth on the basis of local image appearance. Our contributions are: (1) a new scheme for learning and predicting associations between depth and local image appearance using RGB-D data; and (2) the use of this scheme for the initialization of state-of-the-art visual SLAM systems from a single image frame. To the best of our knowledge, this is the first attempt to automatically initialize a SLAM system by associating depth to sensor features through machine learning techniques. We performed a series of tests by making use of the celebrated PTAM system and obtained very promising results. We show successful one-shot initialization examples accomplished by applying our proposed approach to unstructured scene environments.

This paper presents a novel scheme to efficiently secure variable length coded (VLC) multimedia bit streams, such as H.264. The proposed scheme employs code word diffusion and content based shuffling techniques to achieve security. Specifically, it is a combination of a highly secure random number generator based on chaotic maps and a low computation complexity block shuffling procedures. The scheme is ubiquitous to the end users and can be deployed at any node in the network, without requiring any decoding. Its power efficiency and scalability is demonstrated on mobile as well as fixed platforms.

We propose a new collaborative clock offset estimation scheme between two nodes in the Internet using independent one-way offset estimations. Our proposal (different than current schemes in the literature) is intended to provide a fast and accurate clock offset estimation in approximately [Round-Trip Time (RTT)+40]ms. The scheme sends a group of 5 probes in the the forward and reverse paths, and models the One-way Transit Time (OTT) by a Gamma distribution (with parameters adapted to actual path condition) to estimate the minimum distribution value (or long-term minimum OTT value). End nodes exchange their corresponding minimum distribution values to get an improved final clock offset estimate, which takes into account the network path asymmetries. We show that our scheme provides a faster clock offset estimation with lower RMSE and superior stability than NTP and current NTP-like state of the art methodologies in the literature.

This paper presents a geometric error concealment scheme for DCT-based image/video data based on the Bezier polynomials (BP). The proposed scheme makes use of the average edge direction and local curvature (extracted from healthy blocks around the damaged block) as boundary conditions to create an n-degree BP. The resulting BP is then used to interpolate the edges of the lost block, as well as to directionally reconstruct its low frequency data. When multiple edges are identified crossing the lost area, a cost function defined in terms of the local and global edge curvatures is used to find the best match for each missing edge. Experimental results show that our approach provides near perfect reconstruction and excellent subjective quality of the restored data, outperforming current linear interpolation schemes in the literature

Interleaving schemes have proven to be an important mechanism in reducing the effect of network errors on image transmission. Current interleaving schemes spatially decorrelate neighboring image blocks by putting them into packets that are a long way apart from each other in the transmission sequence. Most of the existing schemes, while achieving good performance on random packet losses, do not work well in the case of bursty packet losses. In this paper, we propose two interleaving schemes where the decorrelation process is applied not only at the block level but also at the pixel or coefficient level (i.e. on the information within the blocks) in the compressed domain. The decorrelation is achieved via a k-way shuffle of pixels among the blocks and by using the spatial properties of different space-filling curves, where k depends on the total number of blocks in the image. Our results show that in spite of the decorrelation of coefficients in the triangular interleaving scheme, the compression ratio is within 92% to -98% of JPEG standard compression. We also show that, when compared with existing interleaving schemes, our techniques provide improved image quality and a low mean square error in the transmitted images in the presence of random as well as bursty packet losses in networks.