灰度烟支图像的二值化算法

A Binarization Algorithm for Gray Cigarette Images

在烟支计数的实际应用中,为了后续的计数,需要对灰度烟支图像二值化。由于烟支图像中烟支小而多,烟支之间的间隔小,使用常用的阈值选取算法来二值化烟支图像并不能取得好的效果。本文根据Pun的最大熵算法提出了一种改进的局部最大熵均值综合阈值选取方法。本文的方法把整个烟支图像分成小区域,在每一个区域中综合考虑最大熵和区域灰度均值来得到最优的阈值。实验结果表明了此算法的有效性。

In cigarette counting applications,a binarization algorithm is needed for the following counting process.But the existing binarization algorithms are not very effective because cigarettes are small and the areas between cigarettes are also small.In this paper,an improved local maximum entropy average thresholding method is implemented according to the maximum entropy method presented by Pun.The improved algorithm divides the whole image into areas and selects the optimum threshold for each area according to the maximum entropy threshold and mean.The experimental results prove the validity of the algorithm.