摘要
图像细化处理在字符识别、指纹识别等模式识别中有着重要的应用。本研究针对图像并行细化算法中存在细化后信息丢失及斜线细化存在冗余像素的问题,提出一种改进的图像并行细化算法。针对斜线细化中出现的问题,在ZS算法和EPTA算法的基础上构造了一组可更好地消除冗余像素的消除模板和避免信息丢失的保留模板;针对二像素直线细化过程中信息丢失的问题,在ZS算法的基础上构造了一组保留模板;针对对称图像细化过程中图像出现不对称性的问题,构造了一组对称的保留模板。实验表明,本研究提出的改进算法能很好地解决图像并行细化过程中产生的信息丢失及像素冗余等问题,使图像完全细化,骨架保留完整。
Image thinning process is very important for image recognition. According to the problems of parallel thinning algorithms, such as information loss and the redundant pixels exiting in slash thinning, an improved parallel thinning algorithm was proposed in this study. A set of elimination templates and reservation templates were constructed to eliminate the redundant pixels and avoid the information loss based on ZS algorithm and EPTA algorithm. Aiming at the loss of information of the two pixels linear, a set of reservation templates was proposed based on ZS thinning algorithm. A set of symmetric reservation templates was proposed for the symmetric image. Experimental results showed that the improved algorithm can solve the problems of the information loss and redundant pixels. It can make the image with the complete thinning and skeleton.
出处
《中国印刷与包装研究》
CAS
2014年第1期30-34,共5页
China Printing Materials Market
基金
曲阜师范大学校级科研基金项目--基于多重特征车牌定位算法的研究(No.XJ201204)
国家自然科学基金面上项目--随机非线性切换系统的智能优化控制器设计及性能分析(No.61374004)
关键词
并行细化算法
消除模板
保留模板
二像素直线
Parallel thinning algorithm
Elimination template
Reservation template
Two pixels linear
