Wen-Hui Li

Novel color feature representation and matching technique for content-based image retrieval

Color features of an image are the most widely used features in content-based image retrieval (CBIR) [1][6] systems. Specifically histogram-based algorithms are considered to be effective for color image indexing. Color histogram [2] describes the global distribution of pixels of an image which is insensitive to variations in scale and easy to calculate. However, the high dimensionality of feature vectors results in high computation cost and space cost. In this paper, we mainly focus on color features and propose a novel method named Color Frequency Sequence Difference (CFSD) to express color images, which only has one numerical value in one color channel. The CFSD is combined with information entropy to realize indexing. The novel approach is described in detail and compared with color histogram method presented in the literature. The experiment is finished and shows that the method proposed in this paper is effective and efficient.

Segmentation-Based Image Retrieval

Color features are important to pictures and they are easy to calculate. Therefore, the features are widely used in content-based image retrieval (CBIR)[4][7]. In the meantime, it lacks space information. In this paper, color spaces are analyzed and YUV color space is chosen. Color and texture features are extracted in segmentation block, so there are space information. Major color, major segmentation block, a new kind of color quantization and a new Gray scale co-existing matrixs method are proposed. Our approach is described in detail and compared with other methods presented in the literature to deal with the same problem. The experiments are finished and show that the method in this paper is effective and efficient.

Qualitative spatial relationships cleaning for spatial data mining

In this article, we investigate the problem of preparing qualitative spatial relations before implementing spatial data mining by checking consistency in a constraint network, which includes topological and cardinal directional relations between pairs of spatial objects. We aim to explore potential spatial relations and possible inconsistency among the data of relationships for enforcing the correctness of spatial data mining. This task is carried out through qualitative spatial reasoning method, specifically consistency checking. We try to lay the theoretical foundation for this kind of problem. Instead of using conventional composition tables, we investigate the interactions between topological and cardinal directional relations with the aid of rules. 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. An example is presented to show the utility of these rules.

A fast algorithm for image analogy using particle swarm optimization

This work employs particle swarm optimization (PSO) based texture synthesis approaches, it differs from the pixel-based texture synthesis method, in this way, and the synthesis speed is increased. We apply particle swarm optimization (PSO) to improve the process of searching and matching in patch-based texture synthesis, change the throughout search method of the original algorithm; speed up the process of synthesis without influencing the quality of the image. This algorithm also accomplishes orientation control of matching patches in structure and details through appropriate parameter setting. In our proposed approach, only a single input sample is required to achieve desirable analogy results.

The geometric constraint solving based on memory particle swarm algorithm

Geometric constraint problem is equivalent to the problem of solving a set of nonlinear equations substantially. The constraint problem can be transformed to an optimization problem. PSO is an evolution computing method. It searches the solution space by creating a better next swarm. The new swarm is produced based on the new individuals. Memory particle swarm algorithm is a PSO algorithm that adds a memory influence. We introduce MPSO into geometric constraint solving. The purpose of the added memory feature is to maintain spread and therefore diversity by providing individual specific alternate target points to be used at times instead of the current local best position. The experiment indicates that the algorithm is effective.

Particle swarm optimization-based texture synthesis and texture transfer

In this paper, a fast texture synthesis algorithm-swarm intelligence-based texture synthesis is presented. The algorithm uses an intelligent particle swarm optimization (PSO) algorithm to search the best matching patch from the input sample textures and paste it onto the output texture. After a series of experiments, it is proved that the algorithm is effective to a variety of textures, and high quality textures can be synthesized with a mid-level PC in real-time by using it. Furthermore, we extend it to an application-texture transfer.

A New Technique for Ray Tracing Point-Based Geometry

In this paper a new technique for ray tracing point-based models is presented. Our approach offers a higher ray tracing speed in comparison with previous methods. During pre-process, we add an important attribute to each point for the purpose of ray tracing. During rendering, an algorithm of intersecting a ray with point geometry is demonstrated to get satisfied results. Our approach makes it possible to render high quality ray traced images with global illumination. It performs well especially in the way of boundary representation. We have tested our idea for shadows, reflection and refraction.

The Research on a Novel Geometric Constraint Solver

When transferring the geometric constraint equation group into the optimization model, we need a method to jump out of the local beat solution so that we can find a global best solution. Considering the speed and global capability, we adopt compound particle group optimization algorithm. Particle swarm optimization algorithm is a kind of evolution computation technology based on group intelligence. In all the evolution computations heuristic function should be included to control its ones own characteristic. These parameters are usually correlated with the specific problem and are defined by the users. Suitable parameter choice needs user abundant experience and correct judgment on the information offered by the problem. More important thing is that these heuristic parameters will influence the convergence characteristic of the algorithm. Because of this even experienced users may choose the not appropriate parameter and then make the problem unable to get effective solution. It needs to carry on some research on these parameters more and more. Here we choose the control parameters as an optimization question in the particle swarm algorithm. Thus heuristic function in the PSO can be controlled by the ordinal genetic algorithm and we form the composite particle swarm optimization algorithm. And we use this algorithm into the geometric constraint solving successfully

The Applicaiton of OEA in the Geometric Constraint Solving

In this paper we propose a new optimization algorithm - organizational evolutionary algorithm (OEA) and apply it into the geometric constraint solving. We transfer the geometric constraint problem into a set of nonlinear equations substantially. In OEA the colony is composed of the organizations. Three organizational evolutionary operators - split operator, merging operator and coordinating operator can lead the colony to evolve. These three kinds of operators have different functions in the algorithm. The experiment shows that OEA has good capability in the geometric constraint solving

An Efficient Collision Detection of Complex Deformable Objects Based on Particle Swarm Optimization Algorithm

A technique for performing collision detection between complex deformable objects by using PSO optimizer is demonstrated. This approach provides a more comprehensive way to trade-off accuracy for computation time. Although the swarm can handle temporal coherence and efficiently search through the highly large primitive pair solution space, we combine it with hierarchies to achieve higher culling efficiency and significantly reduce the size of solution space. At last, we give the precision and efficiency evaluation about the algorithm and find it might be a reasonable choice for deformable models in collision detection systems