基于支持向量机和距离度量的管道内表面图像分类方法研究

Classification Based on SVM and Distance Classfication for Inner Wall Anticorrosive Image in Pipe

针对管道内表面图像的分类问题 ,提出了一种将支持向量机和距离度量相结合 ,构成组合分类器的分类方法。分类时先采用距离度量进行前级分类 ,符合条件则给出分类结果 ,否则拒识并转入 SVM分类器进行分类。该方法充分利用了 SVM识别率高和距离度量速度快的优点 ,并且利用距离度量的结果去指导 SVM的训练和测试。实验表明本方法具有较高的效率和识别精度 。

A classification system for inner wall anticorrosive image in pipe is introduced. The system combines support vector machine (SVM) and distance classification into two layer serial classifier. SVM is used as a new technique for pattern recognition with good generalization performance. However, because of use of quadratic programming optimization techniques, the training of SVM is time consuming, especially when the training data set is very large. So we have two classifiers combined. The distance classifier can classify the images and give the final results when the rejecting rule is not satisfied. Otherwise, the distance classifier rejects to classify the input images. These images are fed into SVM for further classification. The algorithm has advantages of SVM and distance classification. Furthermore, it can use the rejected images to train SVM, thus the training is more efficient. Experiments show that the algorithm has high efficiency and low error rate.