基于纹理的图像检索关键技术研究

摘要

Texture is a common existed and inenarrable feature in image. The extraction and recognition of texture information is the most important component in content-based image retrieval research. A method of the extraction and recognition of texture features based on wavelet transforming and singular value decomposition was raised out Calculating with the distance metric of Euclidean spaces, can get the most similar image of sample image.%纹理是图像中普遍存在而又难以描述的特征,纹理特征的提取与识别是基于内容的图像检索研究中的重要组成部分.提出了一种基于小波包变换与奇异值分解的纹理特征提取与识别算法,通过欧氏空间的距离度量的计算,能得到与样本图像最相似的图像.