Xinbo Gao | Xidian University (original) (raw)

Papers by Xinbo Gao

Lecture Notes in Computer Science, 2004

ABSTRACT A novel de-interlacing algorithm based on motion objects is presented in this paper. In ... more ABSTRACT A novel de-interlacing algorithm based on motion objects is presented in this paper. In this algorithm, natural motion objects, not contrived blocks, are considered as the processing cells, which are accurately detected by a new scheme, and whose matching objects are quickly searched by the immune clonal selection algorithm. This novel algorithm integrates many other de-interlacing methods, so it is more adaptive to various complex video sequences. Moreover, it can perform the motion compensation for objects with the translation, rotation as well as the scaling transform. The experimental results illustrate that compared with the block matching method with full search, the proposed algorithm greatly improve the efficiency and performance.

Proceedings. IEEE International Conference on Multimedia and Expo, 2002

A new caption text extraction algorithm that takes full advantage of the temporal information in ... more A new caption text extraction algorithm that takes full advantage of the temporal information in a video sequence is developed. By detecting the (dis)appearance of caption text in a video stream, we first identify video segment that contains the same caption text. Then using the gray-level vector traced across the segment as the feature vector for a pixel point, we can clearly separate a caption pixel from a background pixel for the entire segment.

Neurocomputing, 2014

ABSTRACT Mass classification is one of the key procedures in mammography computer-aided diagnosis... more ABSTRACT Mass classification is one of the key procedures in mammography computer-aided diagnosis (CAD) system, which is widely applied to help improving clinic diagnosis performance. In literature, classical mass classification systems always employ a large number and types of features for discriminating masses. This will produce higher computational complexity. And the incompatibility among various features also may introduce some negative impact to classification accuracy. Furthermore, latent characteristics of masses are seldom considered in the present scheme, which are useful to reveal hidden distribution pattern of masses. For the above purpose, the paper proposes a new mammographic mass classification scheme. Mammograms are detected and segmented first for obtaining region of interests with masses (ROIms). Then Latent Dirichlet Allocation (LDA) is introduced to find hidden topic distribution of ROIms. A special spatial pyramid structure is proposed and incorporated with LDA for capturing latent spatial characteristics of ROIms. For mining latent statistical marginal characteristics of masses, local patches on segmented boundary are extracted to construct a special document for LDA. Finally, all the latent topics will be combined, analyzed and classified by employing the SVM classifier. The experimental results on a dataset in DDSM demonstrate the effectiveness and efficiency of the proposed classification scheme.

2010 International Conference on High Performance Computing & Simulation, 2010

Robust lossless data hiding (LDH) methods have attracted more and more attentions for copyright p... more Robust lossless data hiding (LDH) methods have attracted more and more attentions for copyright protection of multimedia in lossy environment. One of the important requirements of the robust LDH methods is the reversibility, that is, the host images can be recovered without any distortion after the hidden messages are removed. The reversibility is often guaranteed by the embedding model, which affects the performance of the methods greatly. Another requirement is to have a better robustness, which allows the LDH methods to be well adaptable to the lossless and lossy environment, e.g., JPEG compression. In this paper, we firstly categorize the existing robust LDH methods according to the embedding model, and then make a theoretical analysis on the performance in terms of capacity, invisibility and the robustness. Finally, experimental comparisons are carried out to summarize the advantages and disadvantages of each kind of method.

2010 20th International Conference on Pattern Recognition, 2010

The proportion of aurora region to the field of view is an important index to measure the range a... more The proportion of aurora region to the field of view is an important index to measure the range and scale of aurora. A crucial step to obtain the index is to segment aurora region from the background. A simple and efficient aurora image segmentation algorithm is proposed, which is composed of feature representation based on adaptive local binary patterns (ALBP) and aurora region estimation through block threshold. First the ALBP features of sky image are extracted and the threshold is determined. The aurora image to be segmented is then equally divided into detection blocks from which ALBP features are also extracted. Aurora block is estimated through comparison its ALBP features with the threshold. Simple as it is, processing in huge data set is possible. The experiment illustrates the segmentation effect of the proposed method is satisfying from human visual aspect and segmentation accuracy.

Signal Processing, 2006

The problem of data association for target tracking in a cluttered environment is discussed. In o... more The problem of data association for target tracking in a cluttered environment is discussed. In order to deal with the problem of data association for real time target tracking, a novel data association method based on maximum entropy fuzzy clustering is proposed. Firstly, the candidate measurements of each target are clustered with the aid of the modified maximum entropy fuzzy clustering. Then the joint association probabilities are reconstructed by utilizing the fuzzy membership degree of the measurement belonging to the target. At the same time, in order to deal with the uncertainty of the measurements, a new weight assignment is introduced. Moreover, the characteristic of the discrimination factor is analyzed, and the influence of the clutter density on it is also considered, which enables the algorithm eliminate those invalidate measurements and reduce the computational load. Finally, the simulation results show that the proposed algorithms have advantages over the existing ones in terms of efficiency and low computational load.

2007 IEEE International Conference on Image Processing, 2007

A new approach to mass detection in mammography is presented. The main obstacle of building a mas... more A new approach to mass detection in mammography is presented. The main obstacle of building a mass detection system is the similar appearance between masses and density tissues in breast. Hence, the various features of the extracted regions of interest (ROIs) are analyzed by synthesis. Then the support vector machine (SVM), which is designed later to distinguish masses from normal areas, is employed to classify these ROIs exactly. To further improve the performance of SVM, the relevance feedback (RF) is introduced to filter out the false positives. The experimental results illustrate that SVM classifier can effectively detect the mass areas, and the RF-SVM scheme can be efficiently incorporated into this learning framework to further improve detection performance.

Acoustics, Speech, and Signal Processing, 1988. ICASSP-88., 1988 International Conference on

In this manuscript, we propose an automatic sketch synthesis algorithm based on embedded hidden M... more In this manuscript, we propose an automatic sketch synthesis algorithm based on embedded hidden Markov model (E-HMM) and selective ensemble strategy. The E-HMM is used to model the nonlinear relationship between a photo-sketch pair firstly, and then a series of pseudo-sketches, which are generated based on several learned models for a given photo, are integrated together with selective ensemble strategy to synthesize a finer face pseudo-sketch. The experimental results illustrate that the proposed algorithm achieves satisfactory effect of sketch synthesis.

Text detection and recognition for images in multimedia messaging service is a very important tas... more Text detection and recognition for images in multimedia messaging service is a very important task. Since Chinese characters are composed of four kinds of strokes, i.e., horizontal line, top-down vertical line, left-downward slope line and short pausing stroke, we present Gabor filters with scale and direction varied to describe the strokes of Chinese characters for candidate text area extraction. By establishing four sub-neural networks to learn the texture of text area, the learnt classifiers are used to detect candidate text areas. Experimental results show that the proposed approach can improve the accuracy of text detection and the recognition rate of images in multimedia messaging service.

Neurocomputing

This paper aims to address the face recognition problem with a wide variety of views. We proposed... more This paper aims to address the face recognition problem with a wide variety of views. We proposed a tensor subspace analysis and view manifold modeling based multi-view face recognition algorithm by improving the TensorFace based one. Tensor subspace analysis is applied to separate the identity and view information of multi-view face images. To model the nonlinearity in view subspace, a novel view manifold is introduced to TensorFace. Thus, a uniform multi-view face model is achieved to deal with the linearity in identity subspace as well as the nonlinearity in view subspace. Meanwhile, a parameter estimation algorithm is developed to solve the view and identity factors automatically. The new face model yields improved facial recognition rates against the traditional TensorFace based method.

We discuss a new multi-view face recognition method that extends a recently proposed nonlinear te... more We discuss a new multi-view face recognition method that extends a recently proposed nonlinear tensor decomposition technique. We use this technique to provide a generative face model that can deal with both the linearity and nonlinearity in multi-view face images. Particularly, we study the effectiveness of three kinds of view manifold for multi-view face representation, i.e., the concept-driven, data-driven and hybrid data-concept-driven view manifolds. An EM-like algorithm is developed to estimate the identity and view factors iteratively. The new face generative model can successfully recognize face images captured under unseen views, and the experimental results provide the new method is superior to the traditional TensorFace-based algorithm and the view-based PCA method.

A typical news story contains a brief report by the anchor person(s) in the studio, as well as ne... more A typical news story contains a brief report by the anchor person(s) in the studio, as well as news footage in the field. Investigation shows that our recognizer performs better when indexing audio from the studio than that from the field. In order to automatically extract the "reliable" audio segments for speech retrieval, we attempt to detect studio-to-field transitions by means of video parsing. Our research is based on 146 news stories collected from Hong Kong TVB Jade station. Retrieval using the entire audio track gave (average inverse rank) AIR=0.759 while, with the incorporation of video parsing, we performed retrieval based only on the studio recordings, which produced AIR=0.765.

In the field of cluster analysis, objective function based clustering algorithm is one of widely ... more In the field of cluster analysis, objective function based clustering algorithm is one of widely applied methods so far. However, this type of algorithms need the priori knowledge about the cluster number and the type of clustering prototypes, and can only process data sets with the same type of prototypes. Moreover, these algorithms are very sensitive to the initialization and easy to get trap into local optima. To this end, this paper presents a novel clustering method with fuzzy network structure based on limited resource to realize the automation of cluster analysis without priori information. Since the new algorithm introduce fuzzy artificial recognition ball, operation efficiency is greatly improved. By analyzing the neurons of network with minimal spanning tree, one can easily get the cluster number and related classification information. The test results with various data sets illustrate that the novel algorithm achieves much more effective performance on cluster analyzing t...

For the problem of cluster analysis, the objective function based algorithms are popular and wide... more For the problem of cluster analysis, the objective function based algorithms are popular and widely used methods. However, the performance of these algorithms depends upon the priori information about cluster number and cluster prototypes. Moreover, it is only effective for analyzing data set with the same type of cluster prototypes. For this end, this paper presents a novel algorithm based on support vector machine (SVM) for realizing fully unsupervised clustering. The experimental results with various test data sets illustrate the effectiveness of the proposed novel clustering algorithm based on SVM.

The fuzzy c-means (FCM) algorithm is one of the effective methods for fuzzy cluster analysis, whi... more The fuzzy c-means (FCM) algorithm is one of the effective methods for fuzzy cluster analysis, which has been widely used in unsupervised pattern classification. To consider the different contributions of each dimensional feature of the given samples to be classified, this paper presents a novel FCM clustering algorithm based on the weighted feature. With the clustering validity function as a criterion, the proposed algorithm optimizes the weight matrix using an evolutionary strategy and obtains a better result than the traditional one, which enriches the theory of FCM-type algorithms. The test experiment with real data of IRIS demonstrates the effectiveness of the novel algorithm.

Journal of Electronics (China)

Among the available clustering algorithms in data mining, the CLOPE algorithm attracts much more ... more Among the available clustering algorithms in data mining, the CLOPE algorithm attracts much more attention with its high speed and good performance. However, the proper choice of some parameters in the CLOPE algorithm directly affects the validity of the clustering results, which is still an open issue. For this purpose, this paper proposes a fuzzy CLOPE algorithm, and presents a method for the optimal parameter choice by defining a modified partition fuzzy degree as a clustering validity function. The experimental results with real data set illustrate the effectiveness of the proposed fuzzy CLOPE algorithm and optimal parameter choice method based on the modified partition fuzzy degree.

In the field of cluster analysis, most of the available algorithms were designed for small data s... more In the field of cluster analysis, most of the available algorithms were designed for small data sets, which cannot efficiently deal with large scale data set encountered in data mining. However, some sampling-based clustering algorithms for large scale data set cannot achieve ideal result. For this purpose, a FCM-based clustering ensemble algorithm is proposed. Firstly, it performs the atom clustering algorithm on the large data set. Then, randomly select a sample from each atom as representative to reduce the data amount. And the ensemble learning technique is used to improve the clustering performance. For the complex large data sets, the new algorithm has high classification speed and robustness. The experimental results illustrate the effectiveness of the proposed clustering algorithm.

Proceedings / ICIP ... International Conference on Image Processing

A novel face detection tree based on floatboost learning is proposed to accommodate the in-class ... more A novel face detection tree based on floatboost learning is proposed to accommodate the in-class variability of multi-view faces. The tree splitting procedure is realized through dividing face training examples into the optimal sub-clusters using the fuzzy c-means (FCM) algorithm together with a new cluster validity function based on the modified partition fuzzy degree. Then each sub-cluster of face examples is conquered with the floatboost learning to construct branches in the node of the detection tree. During training, the proposed algorithm is much faster than the original detection tree. The experimental results on the CMU and our home-brew test database illustrate that the proposed detection tree is more efficient than the original one while keeping its detection speed