摘要
针对海底质地的特点,利用灰度纹理共生矩阵作为特征参数,K-L变换对海底底质图像进行降维,采用自组织竞争神经网络对图像进行自动分类,对各分类方法精度进行对比。以海底侧扫声纳图像为例,通过实测数据验算,取得理想的效果。
According to the feature of seafloor images, the co-occurrence matrix applied as the feature vectors, to reduce the dimension of images with K-L transform, and to achieve automatic classfication with self-organization competition neural network. The results of side scan sonar image indicate that this method can be well applied in seafloor classification.
出处
《测绘工程》
CSCD
2013年第1期51-54,共4页
Engineering of Surveying and Mapping
关键词
K-L变换
自组织竞争神经网络
共生矩阵
海底底质分类
K-L transform
self-organization competition neural network
co-occurrence matrix
seafloor classification
