基于模糊聚类的沉底小目标声呐图像的识别方法

A Recognizing Method Based on Fuzzy Clustering on the Sonar Image of a Small Target on the Sea Bed

根据沉底小目标声呐图像的特点 ,进行了有效的特征提取 .针对沉底小目标声呐图像类内特征差异性和分散性大的特点 ,用模糊 c-均值聚类算法 (FCM)对每一类目标的学习样本聚类形成多个聚类中心作为此类目标的模板 ,给出了贴近度的计算方法 ,提出了一种分层模糊模式识别的方案 .

Some effective features are extracted from the sonar image of a small target on the sea bed according to its peculiarity. Some clustering centers of the learning samples of every kind targets are obtained by fuzzy c means clustering algorithm (FCM) according to distinct difference and decentralization in the features among one kind sonar images of a small target on the sea bed and the clustering centers are regarded as the templates of the same kind targets. A method of calculating closeness degree is given and a layered scheme of fuzzy pattern recognition is put forward. The feasibility of the method in the paper is proved by the experiments.