模糊聚类无监督算法在图像识别中的应用  

Application of Unsupervised Fuzzy Clustering Algorithm in Image Recognition

作　　者：磨莉[1] 李龙龙[1] 舒蕾[1] MO Li;LI Long-long;SHU Lei(Shaanxi Polytechnic Institute,Xianyang 712000 China)

机构地区：[1]陕西工业职业技术学院

出　　处：《自动化技术与应用》2020年第1期121-124,159,共5页Techniques of Automation and Applications

基　　金：陕西省教育厅科学研究计划项目（编号18JK0062）

摘　　要：本文就模糊C均值聚类算法的优势与缺陷为主要依据,提出了一种模糊聚类无监督算法,切实应用于图像分割。并提出了基于Polysegment快速分析纹理图像的方法明确聚类数目,在此基础上利用模糊聚类无监督算法获取最终分割结果。通过实验结果表明,模糊聚类无监督算法在图像分割中使用所获得的分割结果可以在很大程度避免图像纹理对分割结果的影响,有效分割目标图像与背景图像,精确度较高,而且对不同图像分割的精确性,幅值变化相对稳定,是一种非常科学有效的图像分割法,值得大力推广应用。In this paper,based on the advantages and disadvantages of fuzzy C-means clustering algorithm,an unsupervised fuzzy clustering algorithm is proposed,which is applied to image segmentation.A method of fast texture image analysis based on Polysegment is proposed to determine the number of clusters.On the basis of this method,the final segmentation results are obtained by using unsupervised fuzzy clustering algorithm.The experimental results show that the segmentation results obtained by the fuzzy clustering unsupervised algorithm in image segmentation can greatly avoid the impact of image texture on the segmentation results,and the target image and the background image are effectively segmented with high accuracy.Moreover,the accuracy and amplitude variation of different image segmentation are relatively stable,which is a very scientific and effective image segmentation method,which is worth popularizing and applying.

关 键 词：模糊聚类 无监督算法 图像识别 图像分割 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]