摘要
描述纸币图像的二值化和字符的分割方法。针对纸币图像的特点,提出一种基于最大方差比的图像二值化算法,并利用自适应遗传算法(Adaptive Genetic Algorithm,AGA)得到最优的阈值。该算法的思想是将图像分成两个类,选取类间方差与类内方差的分离度为适应度函数,当分离度取最大值时对应的灰度值为最优的阈值,实验表明,以此阈值对图像分割能快速准确的对纸币图像二值化。最后讨论基于投影法的纸币字符分割的方法,并且取得比较满意的效果。
This paper mainly introduces the algorithm of currency paper images binary and segmentation of characters. It proposes an image binarization algorithm based on maximal variance, and applies Adaptive Genetic Algorithm （AGA） to calculate the best binarization threshold value of the paper currency. The approach of this algorithm is dividing an image object into two classes firstly. Then choose the separation of level between the inter- class variance and intra - class variance as the fitness function. We can obtain the best threshold when the separation of level is maximal. The experiments illustrate the new algorithm can segment the image accurately and quickly. In the end it discusses the segmentation of currency paper characters based on projection and gets satisfying result.
出处
《计算技术与自动化》
2007年第3期92-95,共4页
Computing Technology and Automation
关键词
自适应遗传算法
二值化
最大方差比
阈值
字符分割
adaptive genetic algorithm
binarization
maximal variance
threshold
segmentation
