基于纹理的活动轮廓三维图像分割研究

Research on 3D Image Segmentation with Active Contour Based on Texture

对具高噪声和低对比度三维图像的识别和分割算法进行了研究。基于活动轮廓模型,用Gabor变换提取图像的纹理特征,根据统计学信息假设,通过偏微分方程水平集和窄带方法求解,获得较基本活动轮廓的算法分割更光滑精确的物体轮廓。实验结果表明:改进模型算法的效率和准确度达到预期效果。

The recognization and segmentation algorithm of the 3D images with the characteristics of high noise and low contrast were studied in this paper. The implemented method was put forward based on the active contour model, extracting the texture characteristics with Gabor transformation, integrated with assumption of statistics information. The PDEs were solved by level set and narrow band method to get more smooth and accurate boundary of objects comparing to the basic active contour algorithm. The segmentation results showed that the efficiency and accuracy would be realized as expectation by the improved algorithm.