基于全局及局部特征融合的建筑图像检索策略

Building Image Search Strategy Based on the Fusion of Global and Local Features

提出一个基于建筑外形表现的特征提取及识别策略,即基于掩模子采样的全局特征提取、基于AdaBoost的局部特征提取、各特征贝叶斯子分类器训练及子分类器融合策略.

This paper presents a feature extraction and recognition strategy based on building shape,for the global features,it will drop dimension of overall characteristics based on the sample,and for the local features,it uses AdaBoost algorithm to select the most critical and distinguished part of Construction region,Bayesian classifier training based on characteristics,and all feature fusion.