Infrared Image Segmentation Based on Improved Spotted Hyena Optimizer
-
摘要: 针对斑点鬣狗优化算法(spotted hyena optimizer,SHO)容易陷入局部最优解、求解质量低等缺点,本文提出使用Lévy飞行和单纯形搜索算法改进SHO(spotted hyena optimizer based on simplex method and Lévy flight, Lévy_SM_SHO)。将Lévy_SM_SHO与Lévy飞行斑点鬣狗优化算法(spotted hyena optimizer based on Lévy flight, Lévy_SHO)、单纯形搜索斑点鬣狗优化算法(spotted hyena optimizer based on simplex method, SM_SHO)和SHO在测试函数上结果进行对比,实验证明改进算法能够取得较好的优化结果。并将Lévy_SM_SHO算法用于红外图像阈值分割问题,通过与粒子群算法(particle swarm optimization, PSO)分割结果对比,证明Lévy_SM_SHO算法能够取得较好的阈值分割结果。
-
关键词:
- Lévy飞行 /
- 单纯形搜索算法 /
- Lévy_SM_SHO /
- Lévy_SHO /
- SM_SHO
Abstract: Based on the shortcomings of the spotted hyena optimizer (SHO), falling into a local optimal solution or a low-quality solution is easy. In this study, the Lévy flight and simplex method are proposed to improve the SHO(Lévy_SM_SHO). Comparing Lévy_SM_SHO to Lévy flight spotted hyena optimizer (Lévy_SHO), simplex method spotted hyena optimizer (SM_SHO), and spotted hyena optimizer (SHO) on the test function, the experiment proves that the improved algorithm can achieve better optimization results. Finally, the Lévy_SM_SHO algorithm is applied to the infrared image threshold segmentation problem. By crosschecking the segmentation results with the particle swarm optimization algorithm (PSO), we proved that the Lévy_SM_SHO algorithm can achieve better threshold segmentation results.-
Key words:
- lévy flight /
- simplex search algorithm /
- Lévy_SM_SHO /
- Lévy_SHO /
- SM_SHO
-
表 1 测试函数
Table 1. Test functions
Function Expression Dimension Search range Minimum F5 ${f_{\rm{5}}}{\rm{(}}x{\rm{) = }}\sum\limits_{i{\rm{ = 1}}}^n {{\rm{[100(}}{x_{i{\rm{ + 1}}}} - {x_i}^{\rm{2}}{{\rm{)}}^{\rm{2}}}{\rm{ + (}}{x_i} - {\rm{1}}{{\rm{)}}^{\rm{2}}}{\rm{]}}} $ 30 [-30, 30] 0 F6 ${f_6}(x) = \sum\limits_{i = 1}^n {{{(|{x_i} + 0.5|)}^2}} $ 30 [-100, 100] 0 F13 $\begin{gathered} {f_{13}}(x) = 0.1\{ {\sin ^3}(3{\rm{ \mathsf{ π} }}{x_1}) + \sum\limits_{i = 1}^n {{{({x_i} - 1)}^2}} [1 + {\sin ^2}(3{\rm{ \mathsf{ π} }}{x_i})] \\ \;\;\;\;\;\;\;\;\;\;\; + {({x_n} - 1)^2}[1 + {\sin ^2}(2{\rm{ \mathsf{ π} }}{x_n})]\} + \sum\limits_{i = 1}^n {u({x_i}, 5, 100, 4)} \\ \end{gathered} $
$u({x_i}, a, k, m) = \left\{ {\begin{array}{*{20}{c}} {k{{({x_i} - a)}^m}, \;\;\;{x_i} > a\;\;\;\;\;} \\ {0, \;\;\;\;\;\;\;\; - a \leqslant {x_i} \leqslant a\;\;} \\ {k{{( - {x_i} - a)}^m}, \;\;{x_i} < a\;\;} \end{array}} \right.$30 [-50, 50] 0 F16 ${f_{16}}(x) = 4{x_1}^2 - 2.1{x_1}^4 + \frac{1}{3}{x_1}^6 + {x_1}{x_2} - 4{x_2}^2 + 4{x_2}^4$ 2 [-5, 5] -1.0316 F17 ${f_{17}}(x) = {({x_2} - \frac{{5.1}}{{4{{\rm{ \mathsf{ π} }}^2}}}{x_1}^2 + \frac{5}{{\rm{ \mathsf{ π} }}}{x_1} - 6)^2} + 10(1 - \frac{1}{{8{\rm{ \mathsf{ π} }}}})\cos {x_1} + 10\;\;\;\;\;\;\;\;\;\;\;$ 2 [-5, 5] 0.398 F20 ${f_{20}}(x) = - \sum\limits_{i = 1}^4 {{c_i}} \exp ( - \sum\limits_{j = 1}^3 {{a_i}_j{{({x_j} - {p_{ij}})}^2}} )$ 6 [0,1] -3.3 表 2 4种算法在测试函数上的测试结果
Table 2. Test results of 4 algorithms on test functions
Function Indicators SHO Lévy_SHO SM_SHO Lévy_SM_SHO F5 Maximum 28.9838 29.6218 29 29.0713 Minimum 28.7005 28.7027 28.5028 28.4771 Average 28.8738 28.9371 28.7862 28.7253 Standard 0.1025 0.0799 0.0947 0.1582 F6 Maximum 7.06 7.02 7.50 6.19 Minimum 0.326 0.0421 0.0182 0.003723 Average 5.23 4.710 3.26 3.19 Standard 1.93 2.29 2.92 2.25 F13 Maximum 2.99 3.2709 3 3.0271 Minimum 2.86 0.0046 2.7835 0.0023 Average 2.95 1.8008 2.8493 1.6320 Standard 0.0302 1.1731 0.0559 1.1147 F16 Maximum -0.11 -1.0101 0 -1.0094 Minimum -1.03 -1.0316 -1.0316 -1.0316 Average -0.94 -1.0252 -0.9882 -1.0304 Standard 0.147 0.0057 0.1444 0.0031 F17 Maximum 3.597 0.7162 0.51318 0.4690 Minimum 0.398 0.3982 0.3979 0.3979 Average 0.642 0.4282 0.4060 0.4031 Standard 0.484 0.0528 0.0201 0.0108 F20 Maximum -1.6 -2.0591 -2.7087 -2.9141 Minimum -3.1 -3.0991 -3.3031 -3.3047 Average -2.6 -2.7685 -3.167 -3.1990 Standard 0.3 0.2243 0.145831 0.09829 表 3 基于PSO-Otsu和Lévy_SM_SHO-Otsu的最佳阈值
Table 3. The optimal thresholds based on PSO-Otsu and Lévy_SM_SHO-Otsu algorithms
Infrared image The number of threshold PSO-Otsu Lévy_SM_SHO-Otsu threshold Abnormal temperature distribution in porcelain sleeve of circuit breaker 1 70 70 2 44, 145 46, 143 3 73, 74, 255 70, 76, 255 4 46, 138, 173, 255 46, 73, 144, 255 5 69, 88, 181, 255, 255 46, 54, 132, 139, 255 6 58, 142, 167, 181, 255, 255 53, 65, 76, 144, 255, 255 The circuit breaker still touches the hair to heat 1 63 63 2 51, 147 57, 146 3 44, 72, 255 63, 71, 255 4 38, 80, 123, 255 37, 63, 121, 255 5 58, 70, 135, 150, 255 55, 62, 72, 255, 255 6 54, 71, 97, 186, 246, 255 58, 72, 82, 138, 245, 255 表 4 基于PSO-Otsu和Lévy_SM_SHO-Otsu算法的适应度函数
Table 4. Fitness function based on PSO-Otsu and Lévy_SM_SHO-Otsu algorithms
Infrared image The number of threshold PSO-Otsu Lévy_SM_SHO-Otsu The value of fitness functions Load switch 1 851.7254 851.7254 2 1096.1 1096.2 3 1631.5 1633.5 4 1746.8 1877 5 2096.3 2197.8 6 2255.8 2614.6 Load switch 1 1282.8 1282.8 2 1486.6 1487 3 2305.5 2362.6 4 2521.7 2565.1 5 2763.8 3437.9 6 3182.6 3635.9 表 5 基于PSO-Otsu和Lévy_SM_SHO-Otsu算法的PSNR和SSIM值
Table 5. PSNR and SSIM values based on PSO-Otsu and Lévy_SM_SHO-Otsu algorithms
Infrared image The number of threshold PSO-Otsu Lévy_SM_SHO-Otsu PSNR SSIM PSNR SSIM Load switch 1 18.6778 0.0981 18.6778 0.0981 2 21.0977 0.1753 21.1041 0.1794 3 18.7711 0.0969 18.9324 0.1011 4 21.7095 0.1741 22.3763 0.1806 5 21.3552 0.1266 21.5929 0.1760 6 21.3714 0.1370 21.9312 0.1580 Load switch 1 19.1335 0.1567 19.1335 0.1567 2 20.7128 0.1686 21.2433 0.1720 3 20, 0060 0.1738 20.1006 0.1741 4 22.5955 0.1882 22.6117 0.2747 5 22.6896 0.1892 22.7556 0.1908 6 24.0231 0.1900 24.1233 0.1964 -
[1] Maryam M, Majid H, Fatemeh N. Air condition's PID controller fine-tuning using artificial neural networks and genetic algorithms[J]. Computers, 2018, 7(2): 32. doi: 10.3390/computers7020032 [2] Ayala H V H, Coelho L D S. Tuning of PID controller based on a multiobjective genetic algorithm applied to a robotic manipulator[J]. Expert Systems with Applications, 2012, 39(10): 8968-8974. doi: 10.1016/j.eswa.2012.02.027 [3] Beyer H G, Schwefel H P. Evolution strategies——a comprehensive introduction[J]. Natural Computing, 2002, 1: 3-52. doi: 10.1023/A:1015059928466 [4] ZHU W, DUAN H. Chaotic predator–prey biogeography-based optimization approach for UCAV path planning[J]. Aerospace Science & Technology, 2014, 32(1): 153-161. http://or.nsfc.gov.cn/bitstream/00001903-5/99973/1/1000009426400.pdf [5] Ghosh A, Das S, Chowdhury A, et. al. An improved differential evolution algorithm with fitness-based adaptation of the control parameters[J]. Information Sciences, 2011, 181(18): 3749-3765. doi: 10.1016/j.ins.2011.03.010 [6] 孙云霞, 刘兆刚, 董灵波. 基于模拟退火算法逆转搜索的森林空间经营规划[J]. 林业科学, 2019, 55(11): 52-62. doi: 10.11707/j.1001-7488.20191107SUN Yunxia, LIU Zhaogang, DONG Lingbo. Spatial forest management planning based on reversion search technique of simulated annealing algorithm[J]. Scientia Silvae Sinicae, 2019, 55(11): 52-62 doi: 10.11707/j.1001-7488.20191107 [7] Rashedi E, Nezamabadi-Pour H, Saryazdi S. BGSA: Binary gravitational search algorithm[J]. Natural Computing, 2010, 9(3): 727-745. doi: 10.1007/s11047-009-9175-3 [8] PAN Q K, WANG L, GAO L. A Chaotic Harmony Search Algorithm for the Flow Shop Scheduling Problem with Limited Buffers[M]. Elsevier Science Publishers B. V., 2011. [9] Shaikh N F, Doye D D. An adaptive central force optimization (ACFO) and feed forward back propagation neural network (FFBNN) based iris recognition system[J]. Journal of Intelligent and Fuzzy Systems, 2016, 30(4): 2083-2094. doi: 10.3233/IFS-151921 [10] 金旭旸. 基于莱维飞行的水波优化算法[J]. 科技创新与生产力, 2019(5): 66-68. doi: 10.3969/j.issn.1674-9146.2019.05.066JIN Xuyang. Water wave optimization algorithm based on Lévy flight[J]. Sci-tech Innovation and Productivity, 2019(5): 66-68. doi: 10.3969/j.issn.1674-9146.2019.05.066 [11] HUANG H, YANG X, HAO Z, et al. A novel ACO algorithm with adaptive parameter[C]//International Conference on Intelligent Computing on Lecture Notes in Computer Science, 2006, 4115: 12-21. [12] NIU B, ZHU Y, HU K, et al. A novel particle swarm optimizer using optimal foraging theory[C]//International Conference on Intelligent Computing on Computational Intelligence and Bioinformatics, 2006, 4115: 61-71. [13] Akay B, Karaboga D. A modified Artificial Bee Colony algorithm for real-parameter optimization[J]. Information Sciences, 2012, 192(1): DOI: 10.1016/j.ins.2010.07.015. [14] QIANG Z, LI H, LIU C, et al. A new extreme learning machine optimized by firefly algorithm[C/OL]//Proceedings of the 2013 Sixth International Symposium on Computational Intelligence and Design of IEEE, 2013: https://ieeexplore.ieee.org/search/searchresult.jsp?newsearch=true&queryText=A%20new%20extreme%20learning%20machine%20optimized%20by%20firefly%20algorithm. [15] Mukherjee A, Mukherjee V. A solution to optimal power flow with DC link placement problem using chaotic krill herd algorithm[C]// International Conference on Emerging Technological Trends of IEEE, 2016: DOI: 10.1109/ICETT.2016.7873756 [16] JIANG T, ZHANG C. Application of grey wolf optimization for solving combinatorial problems: job shop and flexible job shop scheduling cases[J]. IEEE Access, 2018, 6: 26231- 26240. doi: 10.1109/ACCESS.2018.2833552 [17] Dhiman G, Kumar V. Spotted hyena optimizer: a novel bio-inspired based metaheuristic technique for engineering applications[J]. Advances in Engineering Software, 2017, 114: 48-70. doi: 10.1016/j.advengsoft.2017.05.014 [18] Dhiman G, Kaur A. Spotted hyena optimizer for solving engineering design problems[C]// 2017 International Conference on Machine Learning and Data Science (MLDS)of IEEE, 2017: 114-119. [19] Dhiman G, Kumar V. Spotted Hyena Optimizer for Solving Complex and Non-linear Constrained Engineering Problems[M]//Harmony Search and Nature Inspired Optimization Algorithms, Advances in Intelligent Systems and Computing, Singapore: Springer Singapore, 2019, 741: 857-867. [20] 贾鹤鸣, 姜子超, 彭晓旭, 等. 基于改进鬣狗优化算法的多阈值彩色图像分割[J]. 计算机应用与软件, 2020, 37(5): 261-267. doi: 10.3969/j.issn.1000-386x.2020.05.045JIA Heming, JIANG Zichao, PENG Xiaoxu, et al. Multi-threshold color image segmentation based on improved hyena optimization algorithm[J]. Computer Applications and Software, 2020, 37(5): 261-267. doi: 10.3969/j.issn.1000-386x.2020.05.045 [21] 孙悦, 何同祥. 基于莱维飞行的改进蚁群算法的PlD参数优化[J]. 仪器仪表用户, 2019, 26(6): 83-85. doi: 10.3969/j.issn.1671-1041.2019.06.024SUN Yue, HE Tongxiang. Optimization of PID parameters based on improved ant colony algorithm for Lévy flight[J]. Electronic Instrumentation Customers, 2019, 26(6): 83-85. doi: 10.3969/j.issn.1671-1041.2019.06.024 [22] 张烈平, 何佳洁, 于滟琳, 等. 基于蚁群算法优化的布谷鸟搜索算法[J]. 微电子学与计算机, 2018, 35(12): 27-32. https://www.cnki.com.cn/Article/CJFDTOTAL-WXYJ201812005.htmZHANG Lieping, HE Jiajie, YU Yanlin, et al. A cuckoo search algorithm based on ant colony algorithm optimization[J]. Microelectronics & Computer, 2018, 35(12): 27-32. https://www.cnki.com.cn/Article/CJFDTOTAL-WXYJ201812005.htm [23] 张新, 李珂, 严大虎, 等. 改进入侵杂草算法求解柔性作业车间调度问题[J]. 系统仿真学报, 2018, 30(11): 446-453. https://www.cnki.com.cn/Article/CJFDTOTAL-XTFZ201811051.htmZHANG Xin, LI Ke, YAN Dahu et al. Improved intrusion weed algorithm for solving flexible job shop scheduling problem[J]. Journal of System Simulation, 2018, 30(11): 446-453. https://www.cnki.com.cn/Article/CJFDTOTAL-XTFZ201811051.htm [24] 赵洪, 李伟鹏, 刘铁军. 基于改进莱维飞行的狼群算法及其在翼型气动优化设计中的应用[J]. 科学技术与工程, 2019, 19(18): 315-323. doi: 10.3969/j.issn.1671-1815.2019.18.048ZHAO Hong, LI Weipeng, LIU Tiejun. An improved Lévy flight based grey wolf optimization algorithm for aerodynamic design problem[J]. Science Technology and Engineering, 2019, 19(18): 315-323. doi: 10.3969/j.issn.1671-1815.2019.18.048 [25] 肖石林. 基于Lévy飞行的树种优化算法及在图像分割中的应用[D]. 南宁: 广西民族大学, 2019.XIAO Shilin. Tree Seed Optimization Algorithm Based on LévyFlight and Its Application in Image Segmentation[D]. Nanning: Guangxi University for Nationalities, 2019. [26] 莫愿斌, 郑巧燕, 马彦追. 单纯形法的布谷鸟搜索算法及其在约束优化问题中的应用[J]. 计算机与应用化学, 2015(2): 213-218. https://www.cnki.com.cn/Article/CJFDTOTAL-JSYH201502019.htmMO Yuanbin, ZHENG Qiaoyan, MA Yanzhui. Cuckoo search based on simplex method and its application on constrained optimization problems[J]. Computers and Applied Chemistry, 2015(2): 213-218. https://www.cnki.com.cn/Article/CJFDTOTAL-JSYH201502019.htm [27] 莫愿斌, 马彦追, 郑巧燕, 等. 单纯形法的改进萤火虫算法及其在非线性方程组求解中的应用[J]. 智能系统学报, 2014(6): 747-755. https://www.cnki.com.cn/Article/CJFDTOTAL-ZNXT201406020.htmMO Yuanbin, MA Yanzhui, ZHENG Qiaoyan, et al. Improved firefly algorithm based on simplex method and its application in solving non-linear equation groups[J]. CAAI Transactions on Intelligent Systems, 2014(6): 747-755. https://www.cnki.com.cn/Article/CJFDTOTAL-ZNXT201406020.htm [28] 张红霞, 罗毅, 师瑞峰. 基于单纯形法的改进型人工鱼群算法[J]. 计算机应用, 2011(5): 1321-1323. https://www.cnki.com.cn/Article/CJFDTOTAL-JSJY201105048.htmZHANG Hongxia, LUO Yi, SHI Ruifeng. Artificial fish swarm algorithm based on simplex method[J]. Journal of Computer Applications, 2011(5): 1321-1323. https://www.cnki.com.cn/Article/CJFDTOTAL-JSJY201105048.htm [29] 肖辉辉. 基于单纯形法和自适应步长的花朵授粉算法[J]. 计算机工程与科学, 2016, 38(10): 2126-2133. doi: 10.3969/j.issn.1007-130X.2016.10.025XIAO Huihui. A flower pollination algorithm based on simplex method and self-adaptive step[J]. Computer Engineering & Science, 2016, 38(10): 2126-2133. doi: 10.3969/j.issn.1007-130X.2016.10.025