基于内容检索的硅藻细胞自动分类

Automatic Classification of Diatom Cells Based on Content Retrieval

对硅藻细胞分类提出了一种基于内容的分类方法。该方法首先获取带外接圆的目标;然后,对其应用纹理主特征直方图,纹理变化度,纹理角特征3种方法提取出目标的特征向量;最后使用欧氏距离进行相似性度量。实验结果表明,该方法具有较好的分类准确率及召回率。

A content-based classification method is proposed for sorting diatom cells .This method first gets the goal with a circumscribed circle and then extracts the feature vector of the target by applying the histogram of main features of the texture ,texture change degree ,texture angle characteristics .Finally , Euclidean distance is used for similarity measure .Experimental results show that the method has good classification precision rate and recall rate .