【摘要】 图象数据分层(Image Data Hierarchy)处理技术在图象数据压缩编码、边缘检测、特征提取以及模式识别中有着广泛的应用.本文将模糊聚类方法引入图象分层中,根据图象的特点,将2×2象素块划分为包含均匀和轮廓部分的七种类型,建立了表征图象特征的相似度矩阵,修正了普通的动态聚类法,提出了λ值的自适应控制方案.最后,给出了对一幅标准“USC Girl”数字图象处理的实例以验证本方案的可行性.更多还原

【Abstract】 Image Data Hierarchy (IDH) is a significant and useful processing method in image data compression coding, edge detection, feature extraction and pattern recognition. A fuzzy clustering method for IDH is introduced in this paper. We find all blocks of 2×2 pels in a digital picture falling into seven modes including homogeneous part and edges. The fuzzy relation matrices of an image are developed and the classic dynamic dustering method is modified. Also, the scheme of adaptively controlling λ-value is presented. At last, the result of processing the "USC Girl"-a standard 256×256 pels digital image shows that the fuzzy clusteriug method will be more useful in IDH fields.更多还原