计算机工程与应用 ›› 2020, Vol. 56 ›› Issue (4): 184-190.DOI: 10.3778/j.issn.1002-8331.1811-0173

• 图形图像处理 • 上一篇    下一篇

基于自适应步长果蝇优化算法图像分割

宋杰,许冰,杨淼中   

  1. 安徽大学 计算机科学与技术学院,合肥 230601
  • 出版日期:2020-02-15 发布日期:2020-03-06

Image Segmentation Based on Adaptive Step Size Fruit Fly Optimization Algorithm

SONG Jie, XU Bing, YANG Miaozhong   

  1. School of Computer Science and Technology, Anhui University, Hefei 230601, China
  • Online:2020-02-15 Published:2020-03-06

摘要:

为了解决大津法抗噪性能不佳和分割效率不足等问题,提出了一种自适应改进步长的果蝇优化算法,并对大津法图像分割阈值进行优化。根据浓度平均值变化率改变步长的果蝇优化算法,对传统的步长进行改进,前期升半柯西分布为指数变量,开始时均匀递增然后呈S状上升自适应增加步长。后期利用柯西分布容易产生远离原点的随机数作为指数的算子进行扰动,根据浓度平均值变化率自适应改变果蝇寻优步长,利于跳出局部最优解。实验证明,改进的算法在收敛速度和寻优精度上都取得了较好的结果,在对图像分割的应用中的效果也较优于其他算法。

关键词: 收敛速度, 寻优精度, 柯西分布, 阈值分割

Abstract:

In order to solve the problems of poor anti-noise performance and insufficient segmentation efficiency of the Dtsu method. An adaptive improved step size fruit fly optimization algorithm is proposed, and the Otsu method image segmentation threshold is optimized. According to the fruit fly optimization algorithm which changes the step size according to the change rate of the average value of the concentration, the traditional step size is improved. The early half-Cauchy distribution is an exponential variable, which starts to increase uniformly and then increases in S-shaped adaptively step size. In the later stage, the Cauchy distribution is easy to generate the random number far away from the origin as the operator of the index. The change of the average value of the concentration corresponds to adaptively change the optimal step size of the fruit fly, which is beneficial to jump out of the local optimal solution. Experiments show that the improved algorithm achieves better results in convergence speed and optimization precision, and it is better than other algorithms in the application of image segmentation.

Key words: convergence speed, optimization accuracy, Cauchy distribution, threshold segmentation