【摘要】 为了获得图像的情感语义信息,提出了一种基于感兴趣区域的图像情感特征提取方法。首先通过眼动仪得到感兴趣区域,然后划分每一幅图像为感兴趣区域和非感兴趣区域;其次,使用了层次分析法来评价感兴趣区域和非感兴趣区域对图像情感的权重值,然后使用灰度共生矩阵及基于感兴趣区域的主颜色提取算法,分别计算求取了感兴趣区域和非感兴趣区域的纹理和主颜色特征,并进而求得了图像情感的整体特征。更多还原

【Abstract】 In order to get the affective semantic information of the image, a feature extraction method of image emotion based on theRegions of Interest(ROI) was proposed in the paper. Firstly, ROI was obtained by the eye tracker. Secondly, Analytic Hierarchy Pro-cess(AHP) was used to evaluate the weight of ROI and non-ROI for the image emotion. Next, Gray level co-occurrence matrix andthe extraction algorithm of domain color were utilized to compute and get the feature of texture and domain color of ROI and non-ROI respectively, and then the whole feature of image emotion was obtained.更多还原