Wenhui Li

Independent components analysis for representation interest point descriptors

This paper presents a new interest point descriptors representation method based on independent components analysis (ICA). The aim of this algorithm is to find a meaningful image subspace and more compact descriptors. Combination the descriptors with an effective interest point detector, the proposed algorithm has a more accurate matching rate besides the robustness towards image deformations. The proposed algorithm first finds the characteristic scale and the location for the interest points using Harris-Laplacian interest point detector. We use Haar wavelet transform on the neighborhood of the interest points and get low frequency gradient feature vectors. Then ICA is used to model the subspace and reduces the dimension of the feature vectors. The experiments show the efficiency of the proposed algorithm.

Intelligent mosaics algorithm of overlapping images

Panoramic Video which uses 360 degree panoramic image is a new approach for composing virtual environment. The panoramic images can be created by “stitching” together overlapping images taken with an ordinary camera. So image mosaics are very important in creating panorama. In this paper, we proposed an intelligent mosaics algorithm. We first use particle swarm optimization (PSO) to find a certain area which contains sufficient objective characters, then we use pattern matching method to search the matching patch in another image and adjust image; at last, the mosaic image is created by a multi-resolution method. Experimental results testy that this algorithm is able to seamlessly stitch two overlapping images automatically.

Interest region-based image retrieval system based on graph-cut segmentation and feature vectors

In this paper, an interest region-based image retrieval system (IRBIR) that combines similarity contributions from interest regions specified by user in images to form a single value for measuring similarity between images is proposed. The interest region-based framework utilizes the segmentation result to capture the higher-level concept of images. A novel image segmentation based on Graph-Cut is proposed for the final result. The segmentation method in this paper is fast and accurate enough for the real-time image retrieval demand than previous region-based methods. Experimental and comparison results, which are performed using a general purpose database containing 2,000 images, are encouraging.

Representation interest point using empirical mode decomposition and independent components analysis

This paper presents a new interest point descriptors representation method based on empirical mode decomposition (EMD) and independent components analysis (ICA). The proposed algorithm first finds the characteristic scale and the location of the interest points using Harris-Laplacian interest point detector. We then apply the Hilbert transform to each component and get the amplitude and the instantaneous frequency as the feature vectors. Then independent components analysis is used to model the image subspace and reduces the dimension of the feature vectors. The aim of this algorithm is to find a meaningful image subspace and more compact descriptors. Combination the proposed descriptors with an effective interest point detector, the proposed algorithm has a more accurate matching rate besides the robustness towards image deformations.

An evolution computation based approach to synthesize video texture

Texture synthesis is one of the hottest areas in computer graphics, computer vision and image processing fields, and video texture synthesis is one subset of it. We bring forward a new method on video texture synthesis, in which evolution computing technique is introduced into the processes of synthesizing videos. In the method, by analyzing and processing a finite source video clip, Infinite video sequences obtained can be played smoothly in vision. Comparing with many existing video texture synthesis algorithms, this method can not only get high-quality video results without complicated pre-processing of source video, but also improve the efficiency of synthesis.

Graph-Based fast image segmentation

In this paper, we describe a fast semi-automatic segmentation algorithm. A nodes aggregation method is proposed for improving the running time and a Graph-Cuts method is used to model the segmentation problem. The whole process is interactive. Once the users specify the interest regions by drawing a few lines, the segmentation process is reliably computed automatically no additional users’ efforts are required. It is convenient and efficient in practical applications. Experiments are given and outputs are encouraging.

New rules for hybrid spatial reasoning

In this article, we investigate the problem of checking consistency in a hybrid formalism, which combines two essential formalisms in qualitative spatial reasoning: topological formalism and cardinal direction formalism. Instead of using conventional composition tables, we investigate the interactions between topological and cardinal directional relations with the aid of rules that are used efficiently in many research fields such as content-based image retrieval. These rules are shown to be sound, i.e. the deductions are logically correct. Based on these rules, an improved constraint propagation algorithm is introduced to enforce the path consistency.

Modeling and refining directional relations based on fuzzy mathematical morphology

In this paper, we investigate the deficiency of Goyal and Egenhofers method for modeling cardinal directional relations between simple regions and provide the computational model based on the concept of mathematical morphology, which can be a complement and refinement of Goyal and Egenhofers model for crisp regions. Based on fuzzy set theory, we extend Goyal and Egenhofers model to handle fuzziness and provide a computational model based on alpha-morphology, which combines fuzzy set theory and mathematical morphology, to refine the fuzzy cardinal directional relations. Then the computational problems are investigated. We also give an example of spatial configuration in 2-dimensional discrete space. The experiment results confirm the cognitive plausibility of our computational models.

Heterogeneous spatial reasoning

In this article, we investigate the problem of checking consistency in a hybrid formalism, which combines two essential formalisms in qualitative spatial reasoning: topological formalism and cardinal direction formalism. Instead of using conventional composition tables, we investigate the interactions between topological and cardinal directional relations with the aid of rules that are used efficiently in many research fields such as content-based image retrieval. These rules are shown to be sound, i.e. the deductions are logically correct. Based on these rules, an improved constraint propagation algorithm is introduced to enforce the path consistency. The results of computational complexity of checking consistency for constraint satisfaction problems based on various subsets of this hybrid formalism are presented at the end of this article.