基于多特征融合的室内场景识别

Indoor Scene Recognition Based on Multiple Feature Fusion

针对室内场景识别,提出将全局特征与局部特征相结合,利用两类特征在空间尺度上的互补特性,获取更全面的场景图像特征。分别采用Gist算法和PHOG算法进行全局和局部的特征提取,并确定以Gist＋i＊PHOG形式进行特征融合。在获得场景图像特征的基础上,引入支持向量机（SVM）进行室内场景图像的识别,利用1-a-1方法实现室内场景多分类。实验结果表明,该方法对于室内典型场景的识别率可以达到60-80%。

Aiming at indoor scene recognition, a more comprehensive feature extraction method is proposed, which combines the complementary property of local features and global features on the spatial scale. Gist algorithm and PHOG algorithm are used to and extract the global features and local features, respectively. Gist＋i＊PHOG style is determined to implement the feature fusion. Based on the scene feature, support vector machine （SVM） is adopted to identify the indoor scene images and 1-a-1 method is used for multi-classification. Experimental results show that the proposed method can obtain the recognition rate of 60-80% for typical indoor scenes.