基于颜色和纹理特征的图像情感语义分类

Image kansei feature classification based on color and texture features

利用图像的低层特征实现了图像高层情感语义(happy和sad)的分类:通过在HSV颜色空间中提取图像的全局颜色特征,并利用黄金分割原理提取位于视觉中心位置主要区域的局部颜色特征,结合二维Gabor小波变换提取全局图像的纹理特征,实现对自然风景图像进行情感特征提取.采用PCA方法对情感特征进行降维,将降维后的特征向量结合BP神经网络,完成情感语义分类检索.

An approach for classifying natural images into emotional categories （happy vs. sad）was proposed. According to the strong relationship between low level features in an image and human emotion, an emotional semantic query model based on image low level feature description was proposed. By extracting the gloable color kansei features in the HSV color space and according to the theory of golden section, a novel approach for describing the most interesting areas＇ color features was proposed, and 2D Gabor wavelet was combined for describing the emotion in texture feature. PCA was used to reduce multidimensional kansei features, and back propagation neural network was employed to map the low level feature vectors to emotional feature space.