摘要
针对传统分割算法难以解决多目标分割等问题,提出了一种改进的一维Kapur熵多阈值分割算法。该算法依据Kapur熵阈值选择原理,应用图像灰度直方图信息,利用迭代合并和选择方法建立口腔图像中的阈值分割模型,解决了图像分割中阈值的自动获取问题和多阈值并行选择问题,实现了口腔图像中牙齿和病灶的分离。形状准则和一致性准则评价方法证明了该算法在抗噪声方面明显优于自适应阈值方法。获得的分割结果较好地保留了图像的灰度信息和边缘信息,为后续的图像分析和诊断工作提供了保证。
Aming at the situation that it is difficult to solve multi-object segment problem by using traditional image segment algorithms, a improved 1 D Kapur threshold algorithm is proposed. Based on Kapur threshold principle, using image histogram information, the threshold segment model of the nonnasality image is set up through iterative merger and choice method. The problems of auto-obtaining of threshold and multli-threshold parallel choice are solved. The tooth and the focus in the nonnasality image are successfully separated. Evaluation methods of form criterion and consistency criterion prove that the algorithm is superior to self-adaptative threshold algorithm. The result primely keeps the grey information and the edge information, which provides guarantee for next work of image analysis and diagnosis.
出处
《控制工程》
CSCD
2007年第B05期96-98,共3页
Control Engineering of China
关键词
熵
图像分割
阈值
形状准则
entropy
image segment
threshold
form criterion
