论文部分内容阅读
特征提取与聚类分析相结合的图像分割方法可以用于近红外显微图像化学信息的快速提取。针对基于主成分分析(PCA)的特征提取运算较为复杂的缺点,提出了一种加权二维主成分分析(W2DPCA)光谱特征提取方法,与模糊C均值(FCM)算法相结合用于近红外显微图像化学分布信息提取。通过片剂的近红外显微图像的仿真实验,验证了W2DPCA-FCM方法的可行性和有效性。实验结果表明,W2DPCA-FCM方法可以减少计算时间、提高聚类精度,是一种有效的红外显微图像分析方法。
The image segmentation method combined with feature extraction and cluster analysis can be used for rapid extraction of near infrared microscopic image chemical information. Aiming at the shortcomings of feature extraction algorithm based on principal component analysis (PCA), a weighted two-dimensional principal component analysis (W2DPCA) spectral feature extraction method is proposed, which is combined with FCM (Fuzzy C-Mean) Micro Image Chemistry Distribution Information Extraction The feasibility and validity of the W2DPCA-FCM method were verified by the simulation experiments of the tablet near-infrared microscopy images. Experimental results show that the W2DPCA-FCM method can reduce the computation time and improve the clustering accuracy. It is an effective infrared microscopic image analysis method.