欢迎访问《图学学报》 分享到：
图学学报

图学学报

• 视觉与图像 • 上一篇    下一篇

基于改进的指数交叉熵和萤火虫群优化的工业CT 图像分割

  

  1. 1. 宿迁学院信息工程学院，江苏 宿迁 223800；
    2. 南京航空航天大学电子信息工程学院，江苏 南京 211106；
    3. 华中科技大学数字制造装备与技术国家重点实验室，湖北 武汉 430074；
    4. 西华大学制造与自动化省高校重点实验室，四川 成都 610039
  • 出版日期:2017-02-28 发布日期:2017-02-22
  • 基金资助:
    国家自然科学基金项目(61573183)；数字制造装备与技术国家重点实验室开放基金项目(DMETKF2014010)；制造与自动化省高校重点实
    验室开放课题(2014)；宿迁市科技计划项目(S201410，Z201529)；江苏省“六大人才高峰”第十二批人才项目(DZXX-049)；宿迁学院科
    研基金项目(2014KY09)

Segmentation of Industrial CT Image Based on Improved Exponential Cross Entropy and Glow-Worm Swarm Optimization

  1. 1. School of Information Engineering, Suqian College, Suqian Jiangsu 223800, China;
    2. College of Electronic and Information Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing Jiangsu 211106, China;
    3. State Key Laboratory of Digital Manufacturing Equipment and Technology, Huazhong University of Science and Technology, Wuhan Hubei 430074, China;
    4. Provincial Key Laboratory of Manufacturing and Automation, Xihua University, Chengdu Sichuan 610039, China
  • Online:2017-02-28 Published:2017-02-22

PDF (PC)

218

被引次数

1

摘要： 为了进一步提高工业CT 图像分割的精确度和运行速度，提出基于灰度-梯度二维
指数交叉熵和混沌萤火虫群优化的阈值图像分割方法。运用最小指数交叉熵进行阈值分割，解
决了Shannon 熵在零点处无定义的问题。采用灰度-梯度二维直方图能更加准确地实现目标和背
景的划分，提高算法的抗噪性。此外，为了更好地进行阈值的全局搜索，利用立方映射生成的
混沌序列来初始化萤火虫的位置；采用基于立方映射的混沌萤火虫群优化算法搜寻最佳的二维
阈值，以进一步提升运算速度。最后，与基于萤火虫算法的二维熵法、基于遗传算法的二维最
小交叉熵法作了比较。实验结果表明，该方法在分割效果和处理速度上有明显优势。

关键词: 图像分割, 工业CT 图像, 阈值选取, 指数交叉熵, 混沌萤火虫群, 立方映射

Abstract: To further improve the segmentation accuracy and processing speed of CT image in
industrial CT detection system, the industrial CT image threshold segmentation was proposed based
on 2-D exponential cross entropy and chaotic glow-worm swarm optimization. By using the
minimum exponential cross entropy for threshold segmentation, the drawback of undefined value at
zero of Shannon entropy was avoided. At the same time, 2-D histogram based on gray-gradient was
taken to partition the object and background precisely in order to improve the anti-noise performance.
In addition, chaotic sequence generated by cube map was used to initiate individual position for easy global searching, and chaotic glow-worm swarm optimization algorithm based cube map was used to
search for 2-D optimal threshold in order to further increase algorithmic speed. Finally, a large
number of experiments on industrial CT images were processed and then the experimental results
were compared with 2-D entropy method based on firefly algorithm and minimum cross entropy
method based on genetic algorithm. The obtained results show that the proposed method has obvious
advantages in segmentation and processing speed.

Key words: image segmentation, industrial CT image, threshold selection, exponential cross entropy;
chaotic glow-worm swarm, cube mapping