【摘要】 该文提出了一种利用自组织(SOM)神经网络进行彩色图像分类,并从原图中分割出感兴趣的目标子图像,进而进行局部形状检索的彩色图像检索方法。与利用全局特征进行检索的方法相比,此方法具有效率高、速度快等特点,并且被检索的图像受噪声污染的影响较小。更多还原

【Abstract】 A way to retrieve color-images is presented in this paper.By this way,the self-organized neural networks are first used to classify color-images into different catalogues by their chromaticity values,and then the sub-images are derived and their spacial features are taken for the local shape retrieval.Compared with the retrieval ways based on spacial features of the whole image,the alternative one is more efficient and less effective by noise of the images.更多还原