基于颜色空间和纹理特征的图像检索

It aims at the shortcoming of the current CBIR systems – they didn’t take into account the spatial information of images in retrieval and proposes a novel method which includes multiple features – color, spatial and texture to index and retrieval images. To reduce the complexity of time and space, this paper gives the method of twice retrieval, that’s to retrieval the most color similar images from the image databases, then extract the spatial and texture features of these images, match them with the sample image by distance measure functions. In this way, the cost of time and space can be reduced sharply.