一种基于小波分析的纹理背景下的物体分割算法

An Algorithm forObject Segmentation under Texture Background Based on Wavelet Transform

基于物体和作为背景的纹理图象在线性赋范空间Ｌ１中的能量分布不同这一特点，给出了一种基于小波分析的特征选取方法，由此得到最佳的特征描述并运用分裂－合并算法进行图象分割。这种算法本质上是从灰度分布来分割图象算法的推广，它不仅适用于纹理背景下的物体分割，而且适用于纹理区域和平滑区域组成的复合图象的分割以及不同纹理间的分割。

According to the difference between the energy distribution of texture image and that of smooth image, a scheme which automatically selects the optimal feature for each pixel using wavelet analysis is proposed, leading to an optimal feature image for image segmentation. This algorithm is an expansion of that by grey feature. It can be used for segmentation of objects from texture background, and images made up of texture regions and smooth ones.