室内外图像混合特征提取及分类

Mixed Feature Extraction and Classification of Indoor and Outdoor Images

室内外图像分类能帮助人们在庞大的图像数据库中寻找有价值的信息.对图像数据库进行有效的分类,可以减少图像查找的工作量.室内、室外图像由于外部光线环境的影响加上室内图像自身的复杂性,仅提取单一特征已经无法满足需求,在分析图像视觉特性的基础上,采用基于颜色、纹理、形状的混合特征及基于K-近邻算法的图像分类方法,针对483幅在自然条件下拍摄的图像进行分类,获得了较高的正确率.

Indoor and outdoor image classification can help people search for valuable information in a massive image database.Image database with effective classification can reduce the workload of image search.Considering the influence of external light environment and the complexity of indoor image itself,only a single feature can not meet the demand.In this paper,based on the analysis of image visual characteristics,the mixed features of color,texture and shape and k-nearest neighbors algorithm（KNN）are applied to classify 483 images taken under natural conditions and a relatively high correct rate is obtained.