神经切片图像中识别定位线断面的研究

Recognition study on positioning line transection of nerve slice image

识别在神经切片图像中的定位线断面是实现神经切片图像配准的前提。在分析了定位线断面特征的基础上,提取了定位线断面的特征集合并计算了先验概率和类概率密度函数,设计了二值数据的贝叶斯分类器,再用该分类器识别神经切片图像中的定位线断面,得到定位线断面中心像素坐标。实验表明,采用二值数据的贝叶斯分类器识别准确率可以达到90.5%,其识别精度与神经网络识别方法相同,但是识别速度显著快于神经网络识别方法。这为后续进行神经切片图像的配准奠定了基础。

It is very important basic of nerve slice image registration to recognize the positioning line transaction in the slice image.On the basis of the feature analyzing of the positioning line transection,this paper obtains the feature set of posi- tion line transaction and calculates the prior probability and probability density fimction.Then the binary Bayes classification is designed and employed to recognize the position line transaction in the slice image.The result of experiment shows that recognizing accuracy reaches 90.5% while employing the method to recognize 200 nerve slice images.The recognition speed of binary Bayes classification is faster than the probabilistic neural network obviously while they have the same recognizing accuracy.