【作者】 陈涛；

【导师】 谢克明；

【作者基本信息】 太原理工大学 ， 控制理论与控制工程， 2009， 硕士

【摘要】 生物免疫系统是一个高度进化的生物系统,它具有高度自适应、高度分布性、自组织等特性。它能够有效识别入侵的抗原并消除抗原,并保持机体的稳定。随着人们对免疫系统研究的进一步深入,免疫系统强大的识别能力引起了许多学者的关注。人工免疫算法是一种受生物免疫系统启发而设计的新型智能优化算法。它结合了问题的先验知识和生物免疫系统的自适应能力,因而具有在信息处理方面有较强的鲁棒性、在求解优化问题时不要求目标函数具有可导性等附加信息、在搜索过程中能更好地收敛到全局最优解等特点,现已被用于机器学习、异常和故障诊断、机器人行为仿真、函数优化、网络入侵检测等多个领域,表现出强大的性能和效率,被人们认为是最具潜力的人工智能算法之一。本文首先对生物免疫系统的一些基本概念、系统组成、功能及原理进行了介绍;简单分析了人工免疫系统的研究内容、研究现状及基本理论;其次,研究和分析了现有的一些典型免疫算法的基本结构和流程。其次,在分析了传统图像增强原理和方法的基础上,针对克隆选择算法的不足进行了讨论及多方面的改进,提出了一种改进的克隆选择算法,并实际编程实现了改进的克隆选择算法。然后,使用一种新的目标函数评价算法的性能,将新的适应度函数结合改进的克隆选择算法进行图像增强。使用此方法可以自适应找出图像归一化的非完全Beta函数的最优参数值,对原始图像进行仿真实现,仿真结果证明其在增强后视觉效果有较大提高。最后,分析了改进的克隆选择算法的搜索速度及参数改变对算法性能的影响,验证了算法的有效性和鲁棒性。并与其他算法的实际结果进行了比较,进一步说明了改进的克隆选择算法的有效性。更多还原

【Abstract】 Biological immune system is a highly parallel adaptive information learning system, which can identify and remove the antigenic eyeliners invading the body. With people’s immune system further in-depth study, the immune system caused by a strong ability to identify a lot of the attention of scholars. Artificial immune algorithm is a kind of new intelligent optimization algorithm which is inspired by biological immune system. Because this algorithm combines the prior knowledge and the adaptive ability of immune system, it has some characteristics as follow : robust in information processing; not requiring derivable additional information of the objective function in solving optimization problem; be able to find better global optimal solution in the process of searching. Now this kind of algorithm has been used in many fields in which showing excellent performance and efficiency, such as machine learning, unconventionality and malfunction diagnosis, simulation of the behavior of robots, control of robots, intrusion detection of networks, function optimization and so on, so it is considered to be one of the most potential intelligent search algorithms.First of all, in this paper, some basic concepts, framework, functions and principles of the biological immune system are introduced. Then the research range, research status and basic theory of the artificial immune system are simply analyzed.Secondly, based on the analysis on classical image enhancement methods and principle, considering the deficiency of the clone selection algorithm, some improvements are made and then an improved clone selection algorithm is proposed. This algorithm is also been realized by programming and a project to search the optimal or suboptimal coefficients. A new objective function is used to evaluate the performance of algorithm, and new adaptive function combined with intelligence optimum algorithm to enhance images. Using the approach, the optimization parameters in the normalized incomplete Beta function of degraded images can be automatically find out and can reason the degraded types of the original image correctly. The simulation results prove that the visual effects of degraded images are highly improved after enhancement.At last, here image enhancement problems are been solved, validating the efficiency of the algorithm. Searching speed of this algorithm and the influence when changing some parameters is discussed, which proved the improved clone selection algorithm is robust. Then the comparison of the results by improved clone selection algorithm to those by other algorithm validated the efficiency of the algorithm once more.更多还原