基于水平集方法的图像修描技术的研究

发布时间：2018-06-05 01:19

本文选题：图像修描 + 水平集　；参考：《西安理工大学》2006年硕士论文

【摘要】： 为了恢复受损的图像，须对其受损区域进行修描，以满足人们对高质量印刷复制技术的需求，本课题针对该类图像中的结构型破损图像设计了一种基于水平集方法的图像修描技术。该技术可以应用到考古、艺术、军事、广告等领域，它既提高了图像修描的质量，同时也改善了图像的视觉效果，尤其对文物保护和历史资料复原具有十分重要的意义。本文总结了图像处理理论在图像修描中的应用，通过分析人类视觉系统中的心脑认知机制，总结出其对图像修描的一些启发性原则，将一般的图像恢复理论应用到图像修复领域，以最佳猜测原理和贝叶斯结构体系为理论框架，，建立了基于偏微分方程和水平集方法的图像修描模型。本文将修描过程视为修描边界曲线的向内演化过程，在修描边界向内收缩的过程中完成对待修描区域的恢复。同时，我们选取待修描区域边界外部的一系列点来修补图像，这样就避免了待修描区域内的点对修描结果的干扰。修描公式中选取了合适的权重，保证了该修描技术与手工修复图画的过程很类似，从而可以达到较好的修描结果。试验结果表明：与现在流行的一些修描算法相比，本文使用的快速步进算法应用比较简单，并且修描效果较好，运行时间很短，特别是在其实用性方面有了很大的提高。在文章的最后，作者总结了本文设计的图像修描技术存在的问题和不足，并对以后的研究方向做出了说明。
[Abstract]:In order to restore the damaged image, it is necessary to repair the damaged area in order to meet the needs of high quality printing and reproduction technology. In this paper, an image repair technique based on the level set method is designed for the structural damaged images of this kind of images. This technique can be applied to archaeology, art, military affairs, advertisement and so on. It not only improves the quality of image repair and drawing, but also improves the visual effect of image, especially for the protection of cultural relics and the restoration of historical data. In this paper, the application of image processing theory in image redrawing is summarized. By analyzing the cognitive mechanism of heart and brain in human visual system, some enlightening principles of image correction are summarized. The general image restoration theory is applied to the field of image restoration. Based on the theory framework of optimal guess principle and Bayesian structure system, an image correction model based on partial differential equation and level set method is established. In this paper, the redrawing process is regarded as the inward evolution of the boundary curve of the trimming boundary, and the restoration of the repair area is completed in the process of the redrawing boundary shrinking inward. At the same time, we choose a series of points outside the boundary of the area to repair the image, thus avoiding the interference of the points in the area. The appropriate weight is chosen in the formula to ensure that the process of repairing the drawing is very similar to that of the manual repair of the drawing, so that a better result can be achieved. The experimental results show that compared with some popular correction algorithms, the fast step algorithm used in this paper is relatively simple in application, and its effect is better, and the running time is very short, especially in its practicability. At the end of the paper, the author summarizes the problems and shortcomings of the image repair technique designed in this paper, and explains the research direction in the future.
【学位授予单位】：西安理工大学
【学位级别】：硕士
【学位授予年份】：2006
【分类号】：TP391.41

【引证文献】