通过对图像固有的颜色、纹理、形状等特征的自动识别,结合相应的降维方法,改进了传统的图像检索技术,为多媒体数据的检索提供了有益的探索。
In the traditional image retrieval technology, people deal with the image by manual indexing, and then use text retrieval technology for keywords. Such manual methods rely on , to a large extent, understanding of people which often ignore many inherent characteristics of the image information. This paper researches color, texture, shape and other characteristics in image to improve the traditional image retrieval which provide a useful exploration for multimedia data retrieval.