基于支持向量机的空白试卷识别方法

通常情况下,很难用试卷扫描图像的像素灰度值来直接区分空白试卷和非空白试卷.应用支持向量机方法可以有效地识别空白试卷.建立了两个二维线性可分的支持向量机,一个是以图像像素灰度值列向量的标准差的最大值和行向量的标准差的最大值为特征的支持向量机1,另外一个是以图像像素灰度值列向量的标准差的标准差和行向量的标准差的标准差为特征的支持向量机2.在实际应用中,大部分空白试卷应用支持向量机1来识别,对个别的位于支持向量机1的分类间隔(margin)内的试卷样本,支持向量机1有可能出现识别错误,在这种情况下,应用支持向量机2作进一步识别.此方法在HSK空白试卷识别中取得了很好的结果.

Recognition of Blank Examination Paper Based on Support Vector Machine

Usually,it is difficult to use image pixel gray values of examination paper to directly distinguish blank examination papers and non blank examination papers.However,by using support vector machine(SVM) method,it can identify blank examination paper effectively.We constructed two two-dimensional linear separable SVM,one was the SVM 1,by taking the maximum of standard deviation which the image pixel gray values of columns vector and rows vector as features.Another was the SVM 2,by taking the maximum of stand...