基于人工免疫的烟草异物图像多阈值分割

Tobacco and Foreign Material Image Segmentation Based on Artificial Immune Algorithm

烟草异物图像分割是图像异物识别的基本任务。为了快速实现烟草异物图像多阈值分割,提出了一种基于人工免疫算法与最大类间方差法的多阈值烟草异物图像自动分割方法。算法首先定义了图像分割目标函数;接着运用人工免疫算法,结合最大类间方差法以及目标函数对图像进行自动分割,并产生最优的多阈值,从而实现图像的多阈值分割。人工免疫算法中,抗原是指最优图像分割目标函数,而抗体是指最优的多阈值。实验证明,方法对烟草异物图像多阈值分割的效果良好,分类清晰。

Tobacco and foreign material image segmentation is one of the most important jobs in tobacco and foreign material recognition. In order to realize the multi - threshold segmentation of tobacco and foreign material image, an automatic image segmentation method based on AIA（ artificial immune algorithm） and maximum between - cluster variance（Otsu） is presented in this paper. First, the objective function for the image segmentation is presented. Then, an artificial immune approach is presented to generate automatically segmentation thresholds. In this approach, the objective function is regarded as antigens, and the segmentation thresholds are regarded as antibodies. Experiments demonstrate the good performance of the proposed method.