基于小波域和活动轮廓模型的纹理图像分割

Texture Segmentation based on wavelet domain and active contour models

提出一种双树复小波域局部二值模式和活动轮廓模型的纹理图像分割方法。该方法首先使用双树复合小波分解纹理图像,然后使用局部二值模式提取纹理特征。利用最大熵准则对纹理特征图像进行选择,活动轮廓模型用于最后的分割。实验结果表明提出的方法对于合成纹理和自然场景数据集,达到了较高的分割精度。

A texture segmentation method is proposed in this paper based on local binary patterns in dual tree complex wavelet transform domain and active contour models. The method first decomposes the texture image with DTCWT. Then the LBP operator is used to extract texture features. The maximum entropy criterion is used for selecting the feature images. Finally the segmentation is obtained by using ACM. The experimental results show that the method can achieve relatively high segmentation accuracy for both synthetic textures and natural scene images.