English

应用科学学报 ›› 2014, Vol. 32 ›› Issue (4): 427-433.doi: 10.3969/j.issn.0255-8297.2014.04.014

• 信号与信息处理 • 上一篇    下一篇

结合模糊熵和遗传算法的双阈值图像分割

郑毅1,2， 郑苹3,4   

  1. 1. 山东工商学院信息与电子工程学院，山东烟台264005
    2. 西安电子科技大学技术物理学院，西安710071
    3. 华中科技大学图像识别与人工智能研究所，武汉430074
    4. 安徽理工大学计算机科学与工程学院，安徽淮南232001
  • 收稿日期:2013-08-08 修回日期:2014-01-22 出版日期:2014-07-31 发布日期:2014-01-22
  • 作者简介:郑毅，博士，讲师，研究方向：图像测量、三维重建、逆向工程和增强现实，E-mail: zhengyi@sdibt.edu.cn
  • 基金资助:

    国家自然科学基金(No.60970105, No.61173173, No.61272430)；山东省自然科学基金(No.ZR2012FL09，No.ZR2013FM015)；山东省住房和城乡建设厅科技项目基金(No.2011YK060)；山东省高等学校科技计划项目基金(No.J14LN02)资助

Dual Thresholding Method Using Fuzzy Entropy and Genetic Algorithm

ZHENG Yi1,2, ZHENG Ping3,4   

  1. 1. School of Information and Electronic Engineering, Shandong Institute of Business and Technology,
    Yantai 264005, Shandong Province, China
    2. School of Technical Physics, Xidian University, Xi’an 710071, China
    3. Institute for Pattern Recognition and Artificial Intelligence, Huazhong University of Science and Technology,
    Wuhan 430074, China
    4. School of Computer Science and Engineering, Anhui University of Science and Technology, Huainan 232001,
    Anhui Province, China
  • Received:2013-08-08 Revised:2014-01-22 Online:2014-07-31 Published:2014-01-22

PDF

980

被引次数

8

摘要： 在火炮身管弯曲度测量系统中，为了能从光靶图像中同时提取标定图案和激光光斑，提出了一种双阈值图像分割方法. 基于模糊数学理论和最大模糊熵判据，把光靶图像中的像素灰度级分为黑、灰和亮3 个模糊子集，用于畸变校正的标定图案的像素灰度级隶属于黑模糊子集，用于测量的激光光斑的像素灰度级隶属于亮模糊子集.使用改进的模糊指数熵作为分类判据，提高了分类准确性. 通过遗传算法确定模糊熵参数的最优组合，降低了计算复杂度，并且最大模糊熵判据仅含有4 个模糊熵参数，减小了搜索空间. 针对光靶图像进行了双阈值分割实验，并与最大类间方差双阈值法、模拟退火模糊熵法和使用未改进的模糊指数熵的遗传模糊熵法进行了比较. 实验结果表明，所提方法能自动而有效地选取双阈值，且分割效果优于其他3 种双阈值分割方法.

关键词: 图像分割, 阈值, 隶属度函数, 模糊熵, 遗传算法

Abstract:  A dual thresholding method is proposed to extract a calibration pattern and a laser spot from a target image simultaneously in a gun barrel camber measurement system. Based on the fuzzy mathematics theoryand a maximum fuzzy entropy criterion, the proposed method can classify the target image into three fuzzysubsets, namely, dark, gray and bright fuzzy subset by their gray levels. Gray levels of the calibration pattern used for distortion correction belong to the dark fuzzy subset, and the ones of the laser spot for measurement belong to the bright fuzzy subset. An improved fuzzy exponential entropy is used as the classification criterion,which can increase classification accuracy. A genetic algorithm is implemented to search for an optimal combination of fuzzy entropy parameters, which has a low computational complexity. There are only four fuzzy entropy parameters in the proposed method, and the search space is small. The proposed method is tested and compared with an Otsu’s dual thresholding method, a dual thresholding method using fuzzy entropy and simulated annealing algorithm, and a dual thresholding method using unimproved fuzzy exponential entropy and genetic algorithm. Experimental results show that the proposed method can determine dual thresholds automatically and efficiently, and has the best segmentation among the tested methods.

Key words: image segmentation, threshold, membership function, fuzzy entropy, genetic algorithm

中图分类号: