摘要
在高光谱遥感图像分类中,需要大量的训练样本对分类器进行训练,然而对样本标记非常困难并且耗时、昂贵。针对样本标记困难的问题,提出了自适应的样本不确定性与代表性相结合的主动学习选择训练样本。样本的不确定性是利用最优标号与次优标号(best vs second-best,BvSB)的方法计算。用期望最大(expectation maximization,EM)聚类计算样本的代表性。然后将样本的不确定性与代表性通过自适应权重相结合,从而选出含信息量最大的未标注样本加入进行人工标注,并加入到训练样本。通过实验表明,此方法性能更加稳定,准确率也有一定的提高。
In hyperspectral remote sensing image classification, needs a large number of training samples to trainclassifi- er, but labeled sanmpes is very difficult ,time-consuming and expensive. Therefor, we proposed aadaptive method com- bined representative samples with uncertainty samples to select samples. We use the active learning based on the best vs second-best（BvSB） for selecting training samples and take advantage of the expectation maximum （Expectation Maximi- zation, EM） cluster to eomputerepresentativeness. Then uncertainty and representative of the samples combined with aadaptive weight to select most informative unlabeled samples for manual labeling, and join the training set for training classifier. Experiments show that our method is more stable performance and accuracy is also improved.
出处
《国外电子测量技术》
2017年第4期17-20,共4页
Foreign Electronic Measurement Technology
关键词
主动学习
样本不确定性与代表性
期望最大聚类
自适应
高光谱图像分类
active learning
representative and uncertainty of samples
expectation maximization
adaptive
hyperspectral image classification
