Biomedical image classification algorithm combining k-means and multidimensional feature analysis
In order to improve the accuracy and speed of biomedical image classification and enhance the work efficiency of medical researchers and clinical workers,this paper proposes a biomedical image classifi-cation algorithm that integrates k-means and multidimensional feature analysis.The algorithm uses the Local binary patterns LBP and the directional gradient histogram HOG to extract two different dimensions of fea-ture information,including texture features and local features,respectively,from medical images,and com-bines the k-means clustering algorithm with this multidimensional feature analysis to achieve high-precision classification of biomedical images.The simulation results conducted on the open biomedical image dataset BreakHis show that in the binary classification experiment,the proposed algorithm has an accuracy of 99.03％,an accuracy of 99.12％,a recall of 98.96％,and an Fl value of 99.04％.In the binary classi-fication experiment,its performance is also relatively ideal,superior to classification algorithms such as SVM,ELM,and ResNet.
k-means clustering algorithmmultidimensional feature analysisimage classificationtexture featureslocal features
万方数据
维普
