摘要
纹理图像千变万化,目前在纹理分类上没有明确的标准。文章借用自然语言中的概念词对自然纹理进行基于概念的分类,把自然纹理分成花纹、条纹、鱼鳞、波纹、斑纹、木纹、裂纹、绒毛、颗粒等九大类别,建立了自然纹理图像库,并用小波变换提取纹理特征,对这些特征进行基于支持向量机的分类。实验结果表明该分类和识别方法准确率高,可以考虑作为基于图像内容检索的一种方法。
Until now, we have no definite discipline to classify texture images because textures vary too much. In this paper, we make use of the natural words to classify natural textures as nine classes, such as flower texture. And then, we found an image database of these natural textures. We extract these images" features through DWT, and apply SupportVector-Machine to recognize them. We acquire much ideal results from our experiments, so it can be considered as a way of Content-Based Image Retrival.
出处
《微电子学与计算机》
CSCD
北大核心
2006年第8期153-155,共3页
Microelectronics & Computer