论文部分内容阅读
提出一种基于多特征组合与优化Bo W模型的影像地物分类新方法。提取影像的SIFT、GIST、颜色、Census和Gabor等多种类型特征,通过实验分析确定最佳特征组合。针对一般K-Means算法没有考虑各个特征值的权重,提出利用自动加权k-Means算法计算不同特征分量的权值,分别对SIFT、GIST、Gabor特征构建了基于权重的影像特征词汇表,采用基于Soft的词汇编码算法进行影像编码,使用SVM算法完成影像分类。通过实验表明方法能有效提高遥感影像分类准确性,并且具有较好的稳定性和鲁棒性。
A new method of image feature classification based on multi-feature combination and optimization Bo W model is proposed. Extract the image SIFT, GIST, color, Census and Gabor and other types of features, through experimental analysis to determine the best combination of features. Aiming at the fact that the K-Means algorithm does not consider the weight of each eigenvalue, a weighted k-Means algorithm is proposed to calculate the weights of different feature components. A weight-based image feature vocabulary is constructed for SIFT, GIST and Gabor features respectively. Soft lexical coding algorithm for image coding, the use of SVM algorithm to complete image classification. Experiments show that the method can effectively improve the classification accuracy of remote sensing images, and has good stability and robustness.