Catálogo de publicaciones - libros

Elastic image registration plays an important role in medical image registration. For elastic image registration based on landmarks of sub-images, optimization algorithm is applied to extract landmarks. But local maxima of similarity measure make optimization difficult to convergence to global maximum. The registration error will lead to location error of landmarks and lead to unexpected elastic transformation results. In this paper, an elastic image registration method using attractive and repulsive particle swarm optimization (ARPSO) is proposed. For each subimage, rigid registration is done using ARPSO. In attractive phase, particles converge to promise regions in the search space. In repulsive phase, particles are repelled each other along opposition directions and new particles are created, which might avoid premature greatly. Next, thin plate spline transformation is used for the elastic interpolation between landmarks. Experiments show that our method does well in the elastic image registration experiments.

Immune Clonal Selection Algorithm (ICSA) is a new intelligent algorithm that can effectively overcome the prematurity and has fast conver-gence speed. An Immune Clonal Selection Scheduling Algorithm (ICSSA) is proposed by applying ICSA to the input-queued packet switch scheduling in this paper. ICSSA is compared with other previous algorithms about two performance measures: the average delay and the maximum throughput of the switch. Closed-form expressions for these measures are derived under uniform i.i.d. Bernoulli, diagonal and bursty traffic model. The experimental results show that better performances can be obtained by ICSSA, and 100% throughput can be guaranteed for these traffic models.

A new approach was proposed as a prototyping direction optimization of points data. Based on perspective theory, a curve surface was built up on the peak point of the produced point lattice of an entity. A lattice grid was used to represent the represent volume of the entity, which was employed in rapidly calculating the represent volume in real time. After analyzing the optimal object functions and strategy, authors adopted the genetic algorithm on selection of operators, cross breeding operators, mutation operators, iteration termination condition and colony scales, etc. The optimization program was set up using Matlab and the optimization was obtained for prototyping direction. At the end of the paper, the aforementioned approach was verified using an actual example, which was further validated on the rapid prototyping machine. The simulation results show that a three-dimensional reconstruction was not necessary based on this proposed points data prototyping direction optimization. The results also indicate that a perfect optimization has been achieved for real time optimization in a space with 360 degree. On the basis of the aforementioned optimization approach, the best position of an entity can be located for rapid prototyping, which can increase prototyping efficiency and reduce the time spending on prototyping. It then can lower the cost.

This study generates a large variety of 3-dimensional maze (3-D maze) problems with a range of through-path lengths by a genetic algorithm. The 3-D mazes consist of 27 cubes containing a T-shaped cavity stacked into a 3× 3 × 3 cube. When the cubes are stacked with the appropriate orientations, a 3-D maze is formed by the cavities. About 2,000 3-D maze problems with through-path lengths from 18 to 54 segments (two segments per cube) were generated by the genetic algorithm using two evaluation functions generating long and short path lengths respectively.

Network random delays directly cause the degradations of networked control systems. A methodology based on model predictive control (MPC) is proposed to overcome the non-deterministic delays in data communication of the network. The general algorithm of dynamic matrix control (DMC) has been improved to make it suitable for network condition. When data packets can not arrive at target nodes in sequence, the predictive value of the system output can be used to take the place of the actual measure value by the controller, and the predictive value of control input will be acted as the required control value which coordinate the whole control system. The experiment results based on Motor Ethernet Control Open Platform (MECOP) show the effectiveness of the real-time networked control strategy.

A Dynamic Clonal Selection Algorithm for the Resource-Constrained Project Scheduling Problem (RCPSP-DCSA) is proposed in this paper. Based on the mechanism of nature immune system and characteristic of project scheduling, the encoding of solution, some operators (including crossover, mutation, deriving and death) and the function of affinity are given. Through a thorough computational study for a standard set of project instances in PSPLIB, the impact of the parameters on the performance of algorithm are analyzed. Experimental results show RCPSP-DCSA has a good performance and it can reach near-optimal solutions in reasonable time.

This paper compares ant colony optimization (ACO) and evolutionary multi-objective optimization (EMO) for the weather avoidance in a free flight environment. The problem involves a number of potentially conflicting objectives such as minimizing deviations, weather avoidance, minimizing distance traveled and hard constraints like aircraft performance. Therefore, we modeled the problem as a multi-objective problem with the aim of finding a set of non dominated solutions. This approach is expected to provide pilots the additional degree of freedom necessary for self optimized route planning in Free Flight. Experiments were conducted on a high fidelity air traffic simulator and results indicate that the ACO approach is better suited for this problem, due to its ability to generate solutions in early iterations as well as building better quality non dominated solutions over time.

The simulation of land combat operations is a complex task. The space of possibilities is exponential and the performance criteria are usually in conflict; thus finding a sweet spot in this complex search space is a hard task. This paper focuses on the effect of population size and mutation rate on the performance of NSGA–II, as the evolutionary multiobjective optimization technique, to decide on the composition of forces using a complex land combat multi-agent scenario planning tool.

The electroencephalogram (EEG) machine is the most influential tool in the diagnosis of epilepsy, which is one of the most common neurological disorders. In this paper, a new seizure detection approach, which combined the genetic algorithm (GA) and the support vector machine (SVM), is proposed to improve visual inspection of EEG recordings. Genetic operations are utilized to optimize the performance of SVM classifier, which includes three aspects: feature subset selection, channel subset selection and parameter optimization of SVM. These optimization operations are performed simultaneously during the training process. The epileptic EEG data acquired from hospital are divided into two parts of training set and testing set. The results from the test on EEG data show that the method may more effectively recognize the spike and sharp transients from the EEG recording of epileptic patients than those without using optimal operations.

The tracking of articulated body in images sequences is a challenging problem due to complexity and high dimensionality of the configuration space. In this paper, we propose a new algorithm to combine Artificial Immune and particle filter for articulated body motion tracking, fusing the strengths of both approaches. Compared with previous optimization based particle filter, our method overcomes the disadvantages of inefficiency by incorporating artificial immune algorithm into particle filter. Evaluations on MOCAP dataset show that immune particle filter algorithm performs better than anneal particle filter.