一种基于数学形态学的图象多尺度分析方法的研究

提出了一种基于数学形态学的图象多尺度分析方法。利用数学形态学在描述图象形态特征方面的优势，通过改变结构基元的尺度，提取图象在不同尺度下的有用信息。该方法将基于二值图象的集合处理的数学形态学方法，推广至基于灰度形态处理的数学形态学方法，提出了广义的数学形态学形态变换算子，拓宽了数学形态学理论在图象处理领域的应用范围。本文给出了人造的实验图象和实际的景物图象的实验结果，提取出图象在不同尺度下各个层次的目标和特征，证明该方法的有效性。

Multiscale Analysis of Images Based on Mathematical Morphology

Presents an mathematical morphology method to implement the multiscale analysis of images. The most critical work in this paper is to take advantage of morphology method to extract image objects of different sizes. Multiscale primitives are defined to be in agreement with the spatial size of image objects to implement the multiscale analysis. Moreover, the set processimg morphology method of graytone images is developed to greyscale morphology method and a extensive mophological operator is defined. All of them expand the application of morphology theory in image processing. Experimental results of an artificial image and a scene image prove the validity of this method.