CAJ | 학술논문

针对SVM在处理具有样本集规模大的ECT系统数据时，存在ECT图像重建的成像精度不高和速度慢的问题，采用了选择分块支持向量机CSSVM算法。将ECT系统样本数据构成列数固定的样本矩阵，每个样本作为样本矩阵的行，66个电容值和66个敏感度值作为矩阵的列。该算法将大样本矩阵按照某一成像单元进行选择性分块，并形成多个小样本矩阵，再分别采用SVM算法进行训练和预测，将各个成像单元组合成像。数值实验证明，使用CSSVM新算法比单独使用SVM算法重建图像具有更高的分类准确率和更短的成像时间。
침대SVM재처리구유양본집규모대적ECT계통수거시，존재ECT도상중건적성상정도불고화속도만적문제，채용료선택분괴지지향량궤CSSVM산법。장ECT계통양본수거구성렬수고정적양본구진，매개양본작위양본구진적행，66개전용치화66개민감도치작위구진적렬。해산법장대양본구진안조모일성상단원진행선택성분괴，병형성다개소양본구진，재분별채용SVM산법진행훈련화예측，장각개성상단원조합성상。수치실험증명，사용CSSVM신산법비단독사용SVM산법중건도상구유경고적분류준학솔화경단적성상시간。
According to Support Vector Machine(SVM)has low training speed and low accuracy to deal with large scale data in Electrical Capacitance Tomography(ECT)system, a new algorithm that combined SVM with the Choice and Segmentation (CS)is presented and it comes into being a new classifier. Data in ECT system composes a data matrix which is fixed matrix column componented of sixty-six capacitance values and sixty-six sensitivity, the samples as its rows.It divides block selectively from large scale samples for one imaging unit. The numerical experiments show that the mixed algorithm can not only improve the accuracy compared to sole SVM, but also shorten time in imaging.