Gao Li-qun

A Novel Approach to Image Enhancement and Thresholding Based on Fuzzy Theory

Image processing has to deal with many ambiguous situations. Fuzzy set theory is a useful mathematical tool for handling the ambiguity or uncertainty. In order to apply the fuzzy theory, selecting the fuzzy region of membership function is a fundamental and important task. In this paper, a new method of membership function based on fuzzy theory by PSO algorithm optimized was proposed by analyzing the deficiencies of traditional enhancement algorithm. A new entropy definition of a fuzzy set was proposed. The new entropy definition of a fuzzy set was not only related to the membership (fuzzy domain) but also related to the probability distribution (space domain), it can respond to the variety of image input information. In addition, by quoting a novel particle swarm optimization (PSO) algorithm to find the optimization parameters for membership. We have employed the new proposed approach to perform image enhancement and thresholding and obtained satisfactory results.

Application of particle swarm optimization to the train scheduling for high-speed passenger railroad planning

This paper presents an approach for solving the train scheduling for high-speed passenger railroad planning problem through the particle swarm optimization (PSO) for the first time. PSO has demonstrated the ability to deal with non-convex, non-linear, integer-mixed optimization problems. In this formulation, the objective function consists of two terms: the variation of inter-departure times for high-speed trains and the total travel time. This combination of terms results in a non-linear objective function. A case on the train scheduling for high-speed passenger railroad planning problem is presented to show the methodologys feasibility and efficiency, compared with the existing optimal planning methods, the search time of the particle swarm optimization method is shorter and the result is close to the ideal solution, simultaneously.

Power system short-term load forecasting based on neural network with artificial immune algorithm

This paper offers one kind of improved artificial immune algorithm which takes different mutation strategy toward different unit that has various quality. This algorithm conducts self-adapt adjustment between mutation rate and crossover rate in order to achieve balance between search accuracy and search efficiency. This paper conducts DAIA-BPNN short-term power load forecast model based on DAIA algorithm. It uses DAIA algorithm to optimize the weight and threshold of BPNN while overcoming the blindness when selecting the weight and threshold of BPNN. The actual calculation example of the short-term power system load forecast shows that the method presented in this paper has higher forecast accuracy and robustness compared with artificial neural networks and regression analysis model.

Improved random walker interactive image segmentation algorithm for texture image segmentation

In order to improve the accuracy of texture image segmentation, we introduce an improved random walker interactive image segmentation algorithm. With some simple user input, we can achieve accurate segmentation of texture image. Firstly, in this paper, we use gabor energy filters to establishes a texture-based similarity weight function expression. Secondly, we build Laplacian matrix to express the adjacency relation between nodes in an undirected graph. Finally, by solving Dirichlet boundary conditions, a high-quality image segmentation result can be achieved. The final experiments demonstrated that the proposed algorithm more accurately describes the texture image structural information, is more suitable for texture image segmentation.

A short-term load forecasting approach based on support vector machine with adaptive particle swarm optimization algorithm

Aiming at the precocious convergence problem of particle swarm optimization algorithm, adaptive particle swarm optimization (APSO) algorithm was presented. In this algorithm, the notion of species was introduced into population diversity measure. The species technique is based on the concept of dividing the population into several species according to their similarity. The inertia weight was nonlinearly adjusted by using population diversity information at each iteration step. Velocity mutation operator and position crossover operator were both introduced and the global performance was clearly improved. The APSO algorithm was adapted to search the optimal parameters of support vector machine (SVM) to increase the accuracy of SVM. A novel short-term load forecasting model based on SVM with APSO algorithm (APSO-SVM) is presented. The proposed model was tested on a certain electricity load forecasting problem. The empirical results illustrated that the new APSO-SVM model outperformed SVM, BPNN and regression model and can successfully identify the optimal values of parameters of SVM with the lowest prediction error values in load forecasting. Therefore, this model is efficient and practical during a short-term load forecasting of electric power system.

Geodesic active contour, inertia and initial speed

A new force field for geodesic active contours, called inertia, is proposed in this paper. Based on analyzing the evolution process of geodesic active contours and constructing extension velocities in level set methods, it is found that geodesic active contours can inherit their evolution velocities from the previous time step and have the tendency to keep their original evolution state through integrating inertia into them. As an useful complementarity to the force field family for geodesic active contours, inertia force is constructed on the base of gradient vector flow external force field. Experimental results reveal that, compared with gradient vector flow geodesic active contours, geodesic active contours integrated with inertia and gradient vector flow can enter into long indentions and special-shape concavities. Furthermore, they are more tolerant toward initial positions under the action of inertia and initial speeds.

Image Segmentation Using Multiscale Gradient Toboggan

Toboggan is an important tool on image segmentation, and the performance of image segmentation based on toboggan mostly independent on the gradient image. A method of image segmentation was presented, which is toboggan segmentation. The gradient image is computed by using the multi-scale morphological gradient operation. The approach of region merging is used after toboggan segmentation. Region merging was carried out according to region area and region homogeneity. It was proved that the method was valid and more accurate than other common algorithms of image segmentation.

A New Method for Tracing by Using Corner Detecting and k-Nearest Neighbor

This paper deals with object tracing using corner detecting and k-Nearest Neighbor. The goal is to determine in successive frame the object which the variable templates. The method of this paper, first used Hassris operator for corner detecting in the image, then based on the theory of K-Nearest Neighbor (KNN) to reduce the corners, Finally, used the remaining corner to construction k-polygon model for tracing. The experiment results indicated that this method can trace the abnormity object, reduce computing time as well as increase scientific property of the algorithm through statistical testing.

Optimal Planning for the Double-Track Train Scheduling Based on Chaotic Particle Swarm Optimization

This paper proposes a multi-objective optimization model for the double-track train scheduling optimal planning problem. In this study, lowering the fuel consumption cost is the measure of satisfaction of the railway company and shortening the total passenger-time is being regarded as the passenger satisfaction criterion. To overcome the drawbacks of conventional mathematical optimization method in arriving at local optimum and dimension disasters, etc., we introduce the chaotic particle swarm optimization (CPSO) technique into the train scheduling for double-track railroad planning for the first time, from which the supreme scheme is generated. A case on the train scheduling optimal planning problem is presented to illustrate the methodology’s feasibility and efficiency, compared with the existing optimal planning methods, the search time of the particle swarm optimization method is shorter.

Decentralized robust output tracking for large-scale systems with time-varying uncertainties

This paper investigates the problem of decentralized robust output tracking control for a class of large-scale uncertain systems. The uncertainties are time-varying and assumed to satisfy the matching conditions for each isolated subsystem.