Wenbing Tao

Color Image Segmentation Based on Mean Shift and Normalized Cuts

In this correspondence, we develop a novel approach that provides effective and robust segmentation of color images. By incorporating the advantages of the mean shift (MS) segmentation and the normalized cut (Ncut) partitioning methods, the proposed method requires low computational complexity and is therefore very feasible for real-time image segmentation processing. It preprocesses an image by using the MS algorithm to form segmented regions that preserve the desirable discontinuity characteristics of the image. The segmented regions are then represented by using the graph structures, and the Ncut method is applied to perform globally optimized clustering. Because the number of the segmented regions is much smaller than that of the image pixels, the proposed method allows a low-dimensional image clustering with significant reduction of the complexity compared to conventional graph-partitioning methods that are directly applied to the image pixels. In addition, the image clustering using the segmented regions, instead of the image pixels, also reduces the sensitivity to noise and results in enhanced image segmentation performance. Furthermore, to avoid some inappropriate partitioning when considering every region as only one graph node, we develop an improved segmentation strategy using multiple child nodes for each region. The superiority of the proposed method is examined and demonstrated through a large number of experiments using color natural scene images.

Object segmentation using ant colony optimization algorithm and fuzzy entropy

In this paper, we investigate the performance of the fuzzy entropy approach when it is applied to the segmentation of infrared objects. Through a number of examples, the performance is compared with those using existing entropy-based object segmentation approaches and the superiority of the fuzzy entropy method is demonstrated. In addition, the ant colony optimization (ACO) is used to obtain the optimal parameters. The experiment results show that, compared with the genetic algorithm (GA), the implementation of the proposed fuzzy entropy method incorporating with the ACO provides improved search performance and requires significantly reduced computations. Therefore, it is suitable for real-time vision applications, such as automatic target recognition (ATR).

Image Thresholding Using Graph Cuts

A novel thresholding algorithm is presented in this paper to improve image segmentation performance at a low computational cost. The proposed algorithm uses a normalized graph-cut measure as thresholding principle to distinguish an object from the background. The weight matrices used in evaluating the graph cuts are based on the gray levels of the image, rather than the commonly used image pixels. For most images, the number of gray levels is much smaller than the number of pixels. Therefore, the proposed algorithm requires much smaller storage space and lower computational complexity than other image segmentation algorithms based on graph cuts. This fact makes the proposed algorithm attractive in various real-time vision applications such as automatic target recognition. Several examples are presented, assessing the superior performance of the proposed thresholding algorithm compared with the existing ones. Numerical results also show that the normalized-cut measure is a better thresholding principle compared with other graph-cut measures, such as average-cut and average-association ones.

Radar-based fall detection exploiting time-frequency features

Falls of the elderly are a major public health concern. In this paper, we develop an effective fall detection algorithm for application in continuous-wave radar systems. The proposed algorithm exploits time-frequency characteristics of the radar Doppler signatures, and the motion events are classified using the joint statistics of three different features. The effectiveness of the proposed technique is verified through measurement data.

A Flexible and Extensible Framework for Web Image Retrieval System

Text-based image search engine and content-based image retrieval (CBIR) have achieved much progress in commercial and academic community respectively. However, few attempts have been conducted to integrate the two techniques for image retrieval in web context. In this paper, based on a novel web image data model, i.e. Fine-Grained Web Image Model (FGWIM), a flexible and extensible framework for web image retrieval is proposed, which incorporates highlevel semantics and low-level visual features of Web images and supports the visual part of MPEG-7 standard. FGWIM model describes the web image data in several levels of abstraction by fine-grained and structured representation, and gives multiple choices at each level, which provides a good flexibility and extensibility for further feature extraction, similarity measurement, integration of semantic and visual features. Based on FGWIM model and the framework, a web image retrieval system prototype is implemented.

VAST: Automatically Combining Keywords and Visual Features for Web Image Retrieval

A large-scale image retrieval system for the WWW, named VAST (VisuAl & SemanTic image search), is presented in this paper. Based on the existing inverted file and visual feature clusters, we form a semantic network on top of the keyword association on the visual feature clusters. The system is able to automatically combine keyword and visual features for retrieval by the semantic network. The combination is automatic, simple, and very fast, which is suitable for large-scale Web dataset. Meanwhile, the retrieval takes advantage of the semantic contents of the images in addition to the low-level features, which remarkably improves the retrieval precision. The experimental results demonstrate the superiority of the system.

A New Image Thresholding Method Based on Graph Cuts

A novel thresholding algorithm is presented to achieve improved image segmentation performance at low computational cost in this paper. The proposed algorithm uses a normalized graph cut measure as the thresholding principle to distinguish an object from the background. The weight matrices used in evaluating the graph cuts are based on the gray levels of an image, rather than the commonly used image pixels. Therefore, the proposed algorithm occupies much smaller storage space and requires much lower computational costs and implementation complexity than other image segmentation algorithms based on graph cuts. This fact makes the proposed algorithm attractive in various real-time vision applications such as automatic target recognition (ATR). A large number of examples are presented to show the superior performance of the proposed thresholding algorithm compared to existing thresholding algorithms.

Layered shape matching and registration: Stochastic sampling with hierarchical graph representation

To automatically register foreground target in cluttered images, we present a novel hierarchical graph representation and a stochastic computing strategy in Bayesian framework. The graph representation, which contains point-(image primitives), seedgraph-, and subgraph- three levels, are built up following the primal sketch theory to capture geometric, topological, and spatial information both in local and global scale. We use two types of bottom-up algorithms for searching matching candidates to generate the point-level and seedgraph-level representations respectively. Then, the Swendsen-Wang Cuts and Gibbs sampling methods are performed for global optimal solution to generate the final subgraph-level representation, where a mixture bending function and a set of topological operators are defined for matching measurement. Experiments with comparison are demonstrated on standard dataset with outperforming results. Results show that our method can work well even with clutter noise and complex background.

Unifying keywords and visual features within one-step search for web image retrieval

The multi-modal characteristics of Web image make it possible to unify keywords and visual features for image retrieval in Web context. Most of the existing methods about the integration of these two features focus on the interactive relevance feedback technique, which needs the user’s interaction (i.e. a two-step interactive search). In this paper, an approach based on association rule and clustering techniques is proposed to unify keywords and visual features in a different manner, which seamlessly implements the integration within one-step search. The proposed approach considers both Query By Keyword (QBK) mode and Query By Example (QBE) mode and need not the user’s interaction. The experiment results show the proposed approach remarkably improve the retrieval performance compared with the pure search only based on keywords or visual features, and achieve a retrieval performance approximate to the two-step interactive search without requiring the user’s additional interaction.