摘要
为了实现卵石分类数字化,利用数字图像技术获取卵石图像边缘信息,基于最小二乘圆度误差概念构建出一种新的描述卵石形状的参数系统。应用模糊数学原理,使用最大树法对从野外采集的卵石样本进行了模糊聚类识别实验;选取不同的聚类模糊度,结果对应不同的聚类细度、不同的模糊分类准度。对其中一种聚类结果,将实物图与几种标准图进行数据对比确定了图形分类特征,并使用多维平面图法进行聚类结果分析,进一步证明了直观判断卵石形状特征参数与模糊聚类关系的可行性。实验表明,该卵石分类法具有一定的实用性和先进性。
To achieve digital pebble classification, this paper used digital image technology, obtained image edge informa- tion, applied least square roundness error theory to build a new parameter system describing pebble shape. Based on the fuzzy theory, it used maximal tree method and did pebble fuzzy clustering to some pebbles samples. Different clustering fuzziness corresponded to different clustering fineness, different fuzzy classification accuracy. For a classification result, it analyzed the compared data between physical image and several standard images to determine image classification feature, and described characters of clustering results by multidimensional ichnography to further prove the feasibility of intuitive judging the relation- ship between pebble shape characteristic parameter and fuzzy clustering. Experiments show that the pebble classification meth- od is fairly practical and advanced.
出处
《计算机应用研究》
CSCD
北大核心
2014年第5期1566-1570,共5页
Application Research of Computers
基金
国家自然科学基金资助项目(50979066
51279124)
关键词
数字图像技术
卵石分类
形状参数
模糊聚类
digital image technology
pebble classification
shape coefficient
fuzzy clustering
