Catálogo de publicaciones - libros

This paper has discussed how to use feature and topology information to compare 3D CAD models represented by polygonal meshes. In this work we propose an enhanced method to compare CAD models based on shape distributions. A topology-preserving simplification method of polygonal meshes was used to simplify CAD model as the pretreatment for generation of sample points. We improved the method of sampling points and a pair of shape functions more sensitive to shape was employed to construct a 2D shape distribution. The experiential results showed that simplification has a positive effort on shape comparison and our method achieved more effective performance than the conventional one.

In this paper, we propose a new technique for extracting salient regions in an image. Identification of salient regions is useful for region/object based image processing. Previous works on salient regions/points typically involve complex detection and are not always reliable in terms of perceptual importance and robustness. This paper presents an efficient salient-region extraction algorithm based on the significance of accumulated wavelet coefficients. The proposed method is robust to common image processing such as compression, filtering, and geometric distortions. Experimental results substantiate the distinguished performance of the proposed method.

In this paper, we focus on the text extraction of image, and propose a new approach for it into two phases: Firstly, for the effective binarization of text region image, instead of performing the binarization in a constant color plane as in the existing methods, our approach adaptively selects the relatively best color plane for the binarization, which uses the text contrast difference among the color planes. Secondly, to remove the noise in the binary image, we consider the color difference between the text strokes and noises, and the color-based clustering is then utilized to remove the noise for the effective text recognition. The experimental result has shown that the proposed approach is better than the existing methods in terms of the performance of text extraction.

This paper presents a robust approach to segmenting text embedded in complex background. Our approach consists of four steps: smart sampling, unsupervised clustering, the Bayesian decision, post-processing. The experimental results show that it works effectively, and is more efficient in removing complex background residues than the popular K-means method.

Peer-to-Peer (P2P) communications have become a popular alternative solution to provide large-scale video-on-demand (VoD) services. Recent approaches are designed for streaming applications in a unicast infrastructure. As the successful deployment of IP broadcast delivery, the system could have a further improvement when broadcasting scheme can be coupled with P2P paradigm. In this paper, we develop a possible solution for building a VoD system using existing broadcasting protocol coupled with cooperative clients in multicast environment. The objective of this work mainly focuses on addressing one design issue in such framework: reliability. An analytical model is developed to determine the minimum number of cooperative clients required for the system. The results showed that 60 peers, each of which has the availability of 0.4, are enough to leverage the workload of the central server up to 95%, when the startup delay of the system is 10 minutes.

We propose a low computing loop filter for reducing the blocking and ringing artifacts for H.264 video coding standard. One-dimensional regularized smoothing function and regularization parameters are newly defined. The experiment result shows that the proposed loop filter has the low computing complexity with similar performance.

A low complexity video de-interlacing algorithm is presented in this paper which is suitable for SIMD (Single Instruction Multiple Data) processors to be used as a pre-/post-processing option in low-cost consumer electronic devices. It is a block-based motion-adaptive technique that converts an interlaced video to progressive, preserving the details in static or low motion areas while leaving no combing artifacts or without introducing any ghosting artifacts. It adapts to the content and chooses one amongst temporal, spatio-temporal or spatial filtering for de-interlacing on a block by block basis analyzing characteristics such as the extent of motion and the correlation across fields. The proposed scheme has been benchmarked against a recent low complexity motion-adaptive algorithm and the performance has been measured across a number of real and synthetic video sequences. The proposed algorithm offers an order of magnitude complexity reduction on SIMD processors over the reference low complexity algorithm while providing a significantly better fidelity towards the progressive source.

In this paper we propose a fast normalized cross correlation (NCC) algorithm for pattern matching based on combining adaptive multilevel partition with the winner update scheme. This winner update scheme is applied in conjunction with an upper bound for the cross correlation derived from Cauchy-Schwarz inequality. To apply the winner update scheme, we partition the summation of cross correlation into different levels with the partition order determined by the gradient energies of the partitioned regions in the template. Thus, this winner update scheme can be employed to skip the unnecessary calculation. Experimental results show the proposed algorithm is very efficient for image matching under different lighting conditions.

The H.264/AVC standard uses new coding tools such as variable block size, multiple reference frames, intra prediction, etc. Using these coding tools, the coding efficiency has been significantly improved. However, the encoder complexity is greatly increased. We focus on the complexity reduction method of intra-mode decision. Our algorithm first restricts selective prediction modes of intra4×4 using a simple preprocessing. The prediction modes of intra4×4 are used for restricting those of the other intra-modes. Simulation results show that the proposed method outperforms other conventional methods.

In this paper, a fast global motion estimation method is proposed for panoramic video coding. This method accelerates the procedure of global motion estimation (GME) during inter-frame prediction by using two techniques: 1) a compact motion model, which uses only three motion parameters to represent accurately the global motion among the successive panoramic frames. 2) A GME filter, which filters the blocks unreliable and contributing less to the GME, based on a threshold method. The experimental results show the proposed fast global motion estimation method manages to speed up the processing of estimating the motion vector field while maintaining the coding performance.