中文摘要：

针对图像情感语义识别中特征提取的问题,提出了一种加权值的图像特征融合算法,并应用于图像情感语义识别。该方法根据不同特征对情感语义的影响不同,在提取出颜色、纹理和形状特征后通过加权融合为新的特征输入量,并用SVM来实现情感语义的识别。实验结果表明,这种算法比单独使用某种图像特征有更高的准确率。

英文摘要：

Because of the semantic gap, we can only extract the image feature to identify indi- rectly the image emotional semantic. In view of the feature extraction problem of image emotional semantic identification, the image feature fusion algorithm with weights was proposed and applied to the identification of image emotional semantic. According to the effects of the extracted color, texture and shape features of image on emotional semantic, the features were weighted and fused into new feature input. SVM was used to achieve emotional semantic identification. This algo-rithm was more accurate than the method that used only one kind of image features in experiments.