中文摘要：

针对既考虑颜色域又考虑空间域的分割算法对小纹理织物图像产生较大分割失真,而只考虑颜色域信息的算法又对大纹理粗糙图像的纱线缝隙不能有效去除的问题,提出了基于小波分解能量的织物分类算法。先判断出织物粗糙与否,再选用相应算法分割。发现由纱线缝隙等造成的较大纹理主要存在于小波变换的二、三级细节图像上,通过计算这2级的分解系数能量和,判断是否属于大纹理织物,再依据二、三级的能量比判定粗糙程度,实现对织物的分类分割。

英文摘要：

When the segmentation algorithm which considers both the spatial information and color space information is applied to the image of small texture fabric,the segmentation often lacks fidelity.And on the other hand,if the segmentation algorithm which only considers color space information is applied to the image of large texture fabric,it often leads to inefficient removal of the gaps between yarns.Therefore,a fabric image classification segmentation algorithm based on wavelet transform coefficient is suggested.Firstly judge whether or not the fabric texture is rough,and then select a corresponding segmentation algorithm.It is found that the texture information mainly exists in the 2nd and 3rd detail images.Based on the decomposition coefficient of energy,the classification of the fabric can be determined from the texture view.And the rough level can also be detected by the energy ratio of 2nd and 3rd detail images.And fabric classification segmentation is thus realized.