集成模糊特征的Snakes模型分割图像纹理

Snakes Model of Integration Fuzzy Feature for Texture Image Segmentation

对纹理图像进行局部一维离散傅里叶变换,把傅里叶变换能反映图像纹理的高频部分的振幅谱作为纹理的特征向量。利用模糊概念定义模糊特征向量,把模糊特征向量集成到Snakes模型。改进的Snakes模型使活动轮廓曲线在曲线内部力场、图像梯度力场和特征向量对区域的模糊隶属度确定大小的曲线法向力场的共同作用下运动。对纹理实验图像和真实图像的进行了分割实验,实验结果说明新的模型能对图像纹理进行有效的分割,模糊特征的融入扩大了Snakes模型分割图像的范围。

The amplitude spectrum of high frequency is taken as feature vector of texture via local one dimension Fourier transform of image. The fuzzy feature vector is defined by using feature vector with membership function. The fuzzy feature vector is integrated with Snakes model. The improved Snakes model makes the active contour curve move under the interaction of curve internal force, image gradient force and normal force of curve. The module of normal force depends on feature vector membership degree. Experimental results show that the new model can segment the image texture effectively, and Snakes model integrated with fuzzy feature can extend the field of image segmentation.