Wu Lingda

Feature analysis and extraction for audio automatic classification

Feature analysis and extraction are the foundation of audio automatic classification. This paper divides audio streams into five classes: silence, noise, pure speech, speech over background sound and music. We present our work on audio feature analysis and extraction on the frame level and clip level. Four new features are proposed, including silence ratio, pitch frequency standard deviation, harmonicity ratio and smooth pitch ratio. We have presented an SVM based approach to classification. The effectiveness of the features is evaluated in experiments. Experiment results show that the features we selected and proposed are rational and effective.

PUMA-SPA: A Phase Unwrapping Method Based on PUMA and Second-Order Polynomial Approximation

This letter focuses on the phase unwrapping algorithm. A state-of-the-art phase unwrapping method called PUMA which is based on the max-flow/min-cut was proposed recently. The proposed method in this letter postprocesses the results of PUMA to improve the unwrapping results. A pointwise local second-order polynomial approximation method is considered to suppress the noise. We estimate the parameters of the polynomial by solving the overdetermined equations and get the solution with the Least Squares Error Fitting. The proposed algorithm synthesizes the unwrapping with the denoising method and is abbreviated as PUMA-SPA. In the denoising step, adaptive local window sizes are selected to compromise the fitting error and the suppression of noise. Experiments show that the proposed method can achieve better results than the method Congruence Operation and Least Squares Fitting (CO-LSF) proposed recently.

Research and implementation of visualization technology for Virtual Battlefield Environment

Understanding the huge data arriving from battlefield and combining it into a single comprehensive view of the battlefield can be extremely difficult, error prone, and time consuming. Virtual battlefield system has been proven to be able to provide situational awareness as well as to support for conducting planning and shaping operations. In this paper, a series of visualization technologies for different branches of virtual battlefield, which was categorized by its spatial property as well as diversity weapon used were presented, respectively. Experimental results were introduced to illustrate how these technologies could be employed to construct a typical Virtual Battlefield Environment (VBE) visualization system.

Indexing structures for content-based retrieval of large image databases: a review

Content-based image retrieval is a focused problem in current multimedia domain. To obtain better searching results more efficiently in some applications, a proper indexing structure is indispensable. This paper reviews the typical indexing structures in content-based image retrieval at first. Then based on the comparison of their different performance, the paper uncovers the problems in those structures and points out the development direction to improve the performance of CBIR in the future.

Semantic event detection in soccer video by integrating multi-features using Bayesian network

Soccer is the most popular game in the world and people are far more interested in the scoring plays in the game. In this paper, we use a Bayesian network to statistically model the scoring event detection based on the recording and editing rules of soccer video. The Bayesian network fuses the five low-level video content cues (evidences) with the graphical model and probability theory. Thus the problem of event detection is converted to the one of statistical pattern classification. And the learning and inference of the Bayesian network are given in the paper. The experimental results indicate that our method is effective and robust.

3D-Parallel Coordinates: Visualization for time varying multidimensional data

Parallel coordinates can visualize multidimensional data efficiently, during to its shortage in displaying time-varying data, we present a new method to add time dimension, which can extend parallel coordinates into 3D space(3D-Parallel Coordinates) consisted of attribute, range and time dimension. So that time-varying multidimensional data can be displayed as polygonal line cluster for recording and analyzing. A technique called clipping shade is taken to highlight datasets around the current time, and results reveal that 3D-Parallel Coordinates can effectively analyze attributes time-varying character.

Network-oriented massive remote-sensing image organization and application

The developing network technology advances the wide application of massive high-resolution remote-sensing. A relative integral massive image data application system refers to the knowledge of data obtaining, organization, management, publication, index and scheduling, exploitation, etc. Based on the massive image management and publishing technology, this paper, making use of high-resolution image as dataset, does a deep research on the massive image data seamless integration and partitioned organization, constructs a image pyramid partition hierarchy model, and designs two kinds of coding and storage ways to provide disparate organization methods, which can meet needs of multipurpose network application.

Tasks for Visual Analytics in Multilayer Networks

Visual analytics has always been an effective analyzing method of structures, dynamics and functions of a wide variety of complex systems. Based on the list of analytic tasks for networks, the fundamental tasks for visual analytics in multilayer networks are introduced, according to the structural multiplexity of multilayer networks, which would motivate further research in visual analytics techniques and contribute to design the visual analytics systems for multilayer networks.

Research on Multi-resolution Isosurface Extraction Method for 3D Scalar Field

Isosurface extraction and rendering of 3D scalar field is an important method in visual analysis of scalar field data. Traditional isosurface extraction algorithm does not take into account the characteristics of the scalar field itself. In this paper, a 3D scalar field multi-resolution isosurface extraction method is proposed to improve the efficiency of isosurface extraction and rendering. This method not only deals the multi-resolution of the 3D scalar field before the isosurface extraction, but also ensures the integrity of the isosurface in the subsequent isosurface extraction process. The experimental results show that the method can greatly improve the efficiency of isosurface extraction and rendering on the basis of less effect on isosurface.

A Phase Unwrapping Method for Large Scale Interferometric Phase Images

Optical interferometry images are widely used in many research areas such as geoscience and remote sensing. However, the phase image which gets from the sensors is modulo-2π, and an important process is to estimate the absolute phase from the observation which is termed phase unwrapping. This paper addresses the absolute phase estimation of large scale images. The proposed phase unwrapping method is patch based which divides the large interferogram into small patches and does the unwrapping based on Markov Random Field for each small patch. This paper also analyzes the two types of errors that may be caused in the combination of the patches and studies how to get rid of these errors. The advantage of the proposal is that the method can be used for the wrapped phase images with discontinuous areas. Experiments in the paper show the efficiency of the proposed algorithm.