Haar变换是正交变换的一种 ,它曾被用在语音识别的特征提取中 ,在其它方面应用也很广泛 .对于选定的一组特征值 ,数据能量集合成一个数值很小的变换系数 ,Haar变换的效率比Fourier变换和Walsh变换的效率要高的多 .本文研究用Haar变换对纺织产品进行特征提取、缺陷检测和分级 . Haar transform is a kind of orthogonal transform, it has been used in speech recognition feature extraction, and it is widely used in other aspects. For a selected set of eigenvalues, the data energy is assembled into a small transform coefficient , The efficiency of Haar transform is much higher than that of Fourier transform and Walsh transform.In this paper, the feature extraction, defect detection and classification of textile products are studied by Haar transform.