【摘要】 针对X线颅颌影像片边缘检测十分困难的问题,提出一种基于多重相关方差的计算多重分形谱的算法.构建一个X线图像的五邻域多重相关空间,定义图像的多重分形谱的概率测度,对数字X线头影侧位片进行多重分形性质的分析.通过图像的多重分形谱与权重因子相关性的分析,进一步判定线形区间及适当的权重因子,得到X线头影片图像的多重分形的线性区间,验证X线头影片图像的多重分形的几何性质,实现对X线头影侧位片的整体轮廓的边缘提取.该算法具有保持和加强图像的局部特性,突出图像的细微处的特点,克服了多重分形谱对噪声敏感的缺点,能有效抑制图像中的噪声.更多还原

【Abstract】 Multi-correlation variance approach based on multifractal used to estimate the multifractal spectrum is introduced.The probability based on X-ray image multi-correlation space is defined that can be used to estimate the multifractal spectrum.The multifractal spectrum property of cephalogram is analyzed and the relationship between the multifractal spectrum and its weight is discussed,so the multifractal linear region of cephalogram is determined.The geometrical properties of its multifractal are verified.The result of the edge detection of cephalogram used the above approach shows that the noise can be repressed effectively,the local property of an image can be remained as much as possible and the fine property can be stressed.更多还原