【作者】 孟玲玲；

【导师】 高隽；

【作者基本信息】 合肥工业大学 ， 信号与信息处理， 2003， 硕士

【摘要】 进化计算是近年来人工智能研究领域内受到人们广泛关注的一个重要方向，也是智能信息处理中的一项重要内容。作为一种基于生物进化原理的优化算法，进化计算与其他优化算法相比，最突出的优点表现在其强大的全局寻优能力上。图像匹配是图像处理、模式识别过程中的一个重要环节。但当匹配模板与待匹配图像之间存在亮度、缩放及角度旋转等差异时，利用传统的匹配算法所得到的匹配结果不尽如人意。本文针对这一问题，提出一种基于进化计算的图像匹配算法，该算法将图像匹配问题看成为一种寻找最优匹配点的全局寻优问题，从而采用进化计算方法来有效解决了该问题。 本文进行的主要工作如下： 1．本文通过实验来对遗传算法和进化策略两种进化算法的适用范围进行了一定的研究与探讨。实验中同时对遗传算法中的二进制编码和实数编码两种编码方式的性能优缺点也进行了相应分析。 2．本文对进化策略中的(μ+λ)和(μ，λ)两种典型的选择方法的性能特点及参数选择方法进行了实验性分析和比较。 3．在上述工作的基础上，本文针对传统图像匹配领域中所存在的问题，提出了一种基于进化计算的图像匹配算法。该算法将图像匹配问题看作为寻求最优匹配点的寻优问题，然后利用进化算法的强大的全局寻优性能来对图像进行匹配。实验结果证明，当待匹配图像和匹配模板之间存在亮度、大小缩放和旋转角度差异的情况下，这种基于进化计算的图像匹配算法能够进行正确匹配。该项工作为进化计算在图像处理领域的进一步应用进行了有益的探索和尝试。更多还原

【Abstract】 Evolutionary Computation is an important branch of artificial intelligence research that is given broad attention in recent years; it is also a main part of intelligent information processing as well. As an optimization algorithm based on the theory of biologic evolution, the most outstanding advantage of Evolutionary Computation is its strong global optimizing capability comparing with other optimization algorithms. Image Matching is a very important part in the course of image processing and pattern recognition. But when there are several kinds of difference between the matching image model and the matched image, such as in lightness, zoom or in angle, the traditional matching algorithms would not be able to obtain satisfying matching results. Aiming at this problem, we put forward an image-matching algorithm based on Evolutionary Algorithms. In our algorithm, we regard the image-matching problem as another kind of optimization problem that is looking for a most suitable matching point in matched image. Then we use Evolutionary Algorithms to solve the problem effectively.Main work of this paper is as follows:1. The paper researches into the most likely application range of Genetic Algorithms and Evolution Strategy. In the experiments, we also analyze the advantages and disadvantages of real coding and binary coding used by Genetic Algorithms.2. Aiming at the two kind of classic selective methods of Evolution Strategy, selective method and selective method, we use experiments tocompare and analyze their propriety and the way to select their parameters, and .3. Based on above work, this paper puts forward an image-matching algorithm based on Evolutionary Computation. In the algorithm, we look the image-matching problem as a kind of optimization problem that finds a most suitable matching point in matched image. Then we use the strong global optimization capability of Evolutionary Computation to solve the complicated matching problem. The results of the experiments show that, when there are different kinds of differences in lightness, zoom and angle between matching model and matched image, this kind of optimization algorithm based on Evolutionary Computation can solve the problem effectively. So our work can be seen as a good try of Evolutionary Computation for more application in the field of image processing.更多还原