论文部分内容阅读
不同的裂缝类型关系到不同养护策略。SVM在解决小样本、非线性、高维度问题时具有较大优势,通过采用不同的SVM分类方法和核函数对常用的数据集中的样本进行分类结果对比,选取了RBF核函数和One-against-All的分类方法。但分类结果仍然满足不了路面养护要求。由于Adaboost选择不同的样本进行训练,改变了训练样本的数据分布。每次迭代都会计算得到一个分类效果最佳的弱分类器及其所在总体分类器中的权重。随着迭代次数的增加,最终由弱分类器迭代生成的强分类器的分类误差最小。提出了SVM-Adaboost分类器动态的对SVM参数进行优化。试验结果表明,应用基于SVM-Adaboost的裂缝分类算法对指定样本进行测试,横向裂缝准确率87.48%,纵向裂缝准确率95.37%,网状裂缝准确率94.9%,块状裂缝准确率89.7%。该方法可以提高组合分类器整体的分类精度。
Different types of cracks relate to different curing strategies. SVM has great advantages in solving small sample, non-linear and high-dimensional problems. By using different SVM classification methods and kernel functions to classify samples in commonly used data sets, the RBF kernel function and One-against- All the classification method. However, the classification results still can not meet the road maintenance requirements. Since Adaboost chose different samples for training, the data distribution of training samples was changed. For each iteration, a weak classifier with the best classification result and the weight in the overall classifier of the classifier are calculated. As the number of iterations increases, the strong classifier eventually generated by the weak classifier iteratively generates the smallest class error. The SVM-Adaboost classifier is proposed to dynamically optimize SVM parameters. The experimental results show that the accuracy of the transverse crack is 87.48%, that of the vertical crack is 95.37%, the accuracy of the mesh crack is 94.9% and the accuracy of the massive crack is 89.7%, using the SVM-Adaboost-based crack classification algorithm. This method can improve the classification accuracy of the combined classifier.