EnFCM Clustering Segmentation Method for Infrared Image of Electrical Equipments Based on Fusion Reconstruction
-
摘要: 红外图像分割是电气设备红外故障诊断的关键环节，而电气设备的不均匀散热、较低的对比度与多源噪声的干扰，会导致目标区域过分割，严重影响分割精度。对此本文提出一种基于融合重构的EnFCM（Enhanced Fuzzy C-Means）聚类电气设备红外图像分割方法。首先对梯度图像进行自适应形态学重建操作，保证算法对噪声图像的分割能力；其次对图像进行显著性检测，将显著图与梯度图融合得到重构后的图像，凸显故障部位的特征，避免过分割；然后对重构后的图像进行分水岭分割获取超像素图像，最后对超像素图像直方图聚类得到分割结果。对电气设备红外图像的实验结果表明：本文算法在电气设备红外图像上能准确分割出故障区域，获取其位置与轮廓，有效改善了过分割现象，在选取的交并比与DICE系数指标对比中，本文方法对比选取的FRFCM、FCM、SFFCM、FCM_SICM、RSSFCA、AFCF平均提升了81%与79%；同时对噪声有较强的鲁棒性，在选取的分割准确率指标对比中，本文方法对比选取的FRFCM、FCM、SFFCM、FCM_SICM、RSSFCA、AFCF平均提升了73%，取得了较优的分割效果。Abstract: Infrared image segmentation plays a pivotal role in diagnosing faults in electrical equipment using infrared imagery. However, uneven heat dissipation, lower contrast, and interference from multiple sources of noise in electrical equipment can lead to over-segmentation of the target region, seriously affecting segmentation accuracy. In this study, we propose an Enhanced Fuzzy C-Means (EnFCM) clustering method based on fusion reconstruction for infrared image segmentation of electrical equipment. First, the gradient image was subjected to an adaptive morphological reconstruction operation to ensure the segmentation ability of the algorithm on noisy images; second, the image was tested for saliency, and the reconstructed image was obtained by fusing the saliency map with the gradient map to highlight the features of the fault site and avoid over-segmentation; then, watershed segmentation was performed on the reconstructed image to obtain the super-pixel image; finally, the histogram clustering of the super-pixel image was obtained by segmentation. The experimental results on the infrared image of electrical equipment show that the algorithm in this paper can accurately segment the fault area on it, obtain its location and contour, and effectively improve the phenomenon of over-segmentation and in the comparison of the selected intersection and concatenation ratio and DICE coefficient indexes, this paper's method improves 81% and 79% on average compared to selected FRFCM, FCM, SFFCM, FCM_SICM, RSSFCA, and AFCF; meanwhile, it is extremely robust to noise, and in the comparison of selected segmentation accuracy indexes, this paper's method achieves segmentation results that are on average 73% superior compared to selected FRFCM, FCM, SFFCM, FCM_SICM, RSSFCA, and AFCF, thus, superior segmentation results were achieved.
-
-
![]()
图 1 基于融合重构的EnFCM聚类分割方法整体框架
Figure 1. Integral framework of EnFCM cluster segmentation method based on fusion reconstruction
![]()
图 2 自适应形态学重建去噪效果。(a)含噪图像；(b)去噪结果
Figure 2. Adaptive morphological gradient reconstruction denoising effect. (a) noisy image; (b) denoising result
![]()
图 3 自适应形态学重建后梯度图像。(a)原梯度图像；(b)重建后的梯度图像
Figure 3. Gradient image after adaptive morphological reconstruction. (a) Original gradient image; (b) reconstructed gradient image
![]()
图 4 FCM方法分割效果。(a)定子匝间短路；(c)转子断条故障；(b)、(d)分割结果
Figure 4. Segmentation effect of FCM method. (a) Stator turn-to-turn short-circuit fault; (c) fault of the rotor break; (b)(d) segmentation results
![]()
图 5 融入了SR前后的梯度图像。(a)融入SR前的梯度；(b)融入SR后的梯度
Figure 5. Gradient images before and after SR are incorporated. (a) Gradient before integration into SR; (b) gradient after integration into SR
![]()
图 6 定子匝间短路（含噪声）。(a)原图；(b)AFCF；(c)FRFCM；(d)FCM；(e)SFFCM；(f)FCM_SICM；(g)RSSFCA；(h)本文算法
Figure 6. Stator turn-to-turn short-circuit fault (Contains noise). (a) Original iamge; (b) AFCF; (c) FRFCM; (d) FCM; (e) SFFCM; (f) FCM_SICM; (g) RSSFCA; (h) proposed
![]()
图 7 转子断条故障（含噪声）。(a)原图；(b)AFCF；(c)FRFCM；(d)FCM；(e)SFFCM；(f)FCM_SICM；(g)RSSFCA；(h)本文算法
Figure 7. Fault of the rotor break (Contains noise). (a) Original iamge; (b) AFCF; (c) FRFCM; (d) FCM; (e) SFFCM; (f) FCM_SICM; (g) RSSFCA; (h) proposed
![]()
图 8 刀闸故障（含噪声）。(a)原图；(b)AFCF；(c)FRFCM；(d)FCM；(e)SFFCM；(f)FCM_SICM；(g)RSSFCA；(h)本文算法
Figure 8. Breaker failure (Contains noise). (a) Original iamge; (b) AFCF; (c) FRFCM; (d) FCM; (e) SFFCM; (f) FCM_SICM; (g) RSSFCA; (h) proposed
![]()
图 9 线夹故障1（含噪声）。(a)原图；(b)AFCF；(c)FRFCM；(d)FCM；(e)SFFCM；(f)FCM_SICM；(g)RSSFCA；(h)本文算法
Figure 9. Contact terminal failure 1 (Contains noise). (a) Original iamge; (b) AFCF; (c) FRFCM; (d) FCM; (e) SFFCM; (f) FCM_SICM; (g) RSSFCA; (h) proposed
![]()
图 10 线夹故障2（含噪声）。(a)原图；(b)AFCF；(c)FRFCM；(d)FCM；(e)SFFCM；(f)FCM_SICM；(g)RSSFCA；(h)本文算法
Figure 10. Contact terminal failure 2 (Contains noise). (a) Original iamge; (b) AFCF; (c) FRFCM; (d) FCM; (e) SFFCM; (f) FCM_SICM; (g) RSSFCA; (h) proposed
![]()
图 11 电抗器故障（含噪声）。(a)原图；(b)AFCF；(c)FRFCM；(d)FCM；(e)SFFCM；(f)FCM_SICM；(g)RSSFCA；(h)本文算法
Figure 11. Inductor failure (Contains noise). (a) Original iamge; (b) AFCF; (c) FRFCM; (d) FCM; (e) SFFCM; (f) FCM_SICM; (g) RSSFCA; (h) proposed
表 1 不同分割方法的IOU与DICE系数值
Table 1 IOU and DICE coefficient values of different segmentation methods
Experimental results Index Methods FRFCM FCM SFFCM FCM_SICM RSSFCA AFCF Proposed Fig.6 IOU 0.5823 0.3658 0.4897 0.5431 0.3256 0.0827 0.9486 DICE 0.6521 0.4046 0.5215 0.6418 0.3391 0.8517 0.9587 Fig.7 IOU 0.7015 0.8019 0.7259 0.5681 0.4033 0.8002 0.9635 DICE 0.7825 0.8693 0.7652 0.5852 0.4231 0.8213 0.9635 Fig.8 IOU 0.8369 0.3845 0.3652 0.6485 0.4988 0.6378 0.9368 DICE 0.8722 0.4316 0.3596 0.6716 0.5021 0.6521 0.9465 Fig.9 IOU 0.8701 0.5864 0.3142 0.4163 0.4374 0.8115 0.9662 DICE 0.7711 0.6364 0.3212 0.4623 0.4516 0.8311 0.9732 Fig.10 IOU 0.8652 0.4562 0.3021 0.6523 0.6881 0.3225 0.9625 DICE 0.8699 0.4214 0.3124 0.7024 0.6557 0.3314 0.9718 Fig.11 IOU 0.8421 0.3022 0.6625 0.7968 0.3225 0.6337 0.9568 DICE 0.8235 0.3158 0.6538 0.8024 0.3126 0.6118 0.9589 
下载: 导出CSV
表 2 不同方法的分割准确率（SA）值
Table 2 Segmentation accuracy (SA) values of different methods
Experimental results FRFCM FCM SFFCM FCM_SICM RSSFCA AFCF Proposed Fig. 6 0.7654 0.4254 0.6535 0.5674 0.4885 0.8005 0.9654 Fig. 7 0.8243 0.7635 0.7224 0.7423 0.7214 0.7968 0.9824 Fig. 8 0.8535 0.4575 0.4635 0.5663 0.8957 0.5671 0.9633 Fig. 9 0.3525 0.4012 0.8026 0.4685 0.7135 0.8213 0.9736 Fig. 10 0.8221 0.4115 0.4021 0.6587 0.5425 0.5732 0.9689 Fig. 11 0.8365 0.4965 0.4215 0.6652 0.4002 0.5235 0.9775
下载: 导出CSV
-
[1] Kim H J, Lee J H, Baek D H, et al. A study on the thermal performance of batteries using thermal imaging and infrared radiation[J]. Journal of Industrial and Engineering Chemistry, 2017, 45: 360-365. DOI: 10.1016/j.jiec.2016.10.003
[2] 赵庆生, 王雨滢, 王旭平, 等. 基于新型阈值选择方法的变电站红外图像分割[J]. 光学学报, 2019, 39(8): 109-116. https://www.cnki.com.cn/Article/CJFDTOTAL-GXXB201908014.htm ZHAO Qingsheng, WANG Yuying, WANG Xuping, et al. Substation infrared image segmentation based on novel threshold selection method[J]. Acta Optica Sinica, 2019, 39(8): 109-116. https://www.cnki.com.cn/Article/CJFDTOTAL-GXXB201908014.htm
[3] 黄志鸿, 洪峰, 黄伟. 形状自适应低秩表示的电力设备热故障诊断方法研究[J]. 红外技术, 2022, 44(8): 870-874. http://hwjs.nvir.cn/article/id/8f0f8a69-4b47-46b4-bcdf-ea623287093f HUANG Zhihong, HONG Feng, HUANG Wei. Shape adaptation low rank representation for thermal fault diagnosis of power equipments[J]. Infrared Technology, 2022, 44(8): 870-874. http://hwjs.nvir.cn/article/id/8f0f8a69-4b47-46b4-bcdf-ea623287093f
[4] 朱惠玲, 牛哲文, 黄克灿, 等. 基于单阶段目标检测算法的变电设备红外图像目标识别及定位[J]. 电力自动化设备, 2021, 41(8): 217-224. https://www.cnki.com.cn/Article/CJFDTOTAL-DLZS202108032.htm ZHU Huiling, NIU Zhewen, HUANG Kecan, et al. Identification and location of infrared image for substation equipment based on single-stage object detection algorithm[J]. Electric Power Automation Equipment, 2021, 41(8): 217-224. https://www.cnki.com.cn/Article/CJFDTOTAL-DLZS202108032.htm
[5] 陈达, 何全才, 迪二镇, 等. 自适应权重的偏微分分割模型在变电设备红外图像中的应用[J]. 红外技术, 2022, 44(2): 179-188. http://hwjs.nvir.cn/article/id/24c18eab-82d8-4403-bb0a-891418242d98 CHEN Da, HE Quancai, DI Erzhen, et al. Application of partial differential segmentation model with adaptive weight in infrared image of substation equipment[J]. Infrared Technology, 2022, 44(2): 179-188. http://hwjs.nvir.cn/article/id/24c18eab-82d8-4403-bb0a-891418242d98
[6] 郭锋, 郑雷, 葛黄徐, 等. 基于相似度阈值模糊聚类的红外区域提取方法[J]. 红外技术, 2022, 44(8): 863-869. http://hwjs.nvir.cn/article/id/2db6a354-f9b9-4fd3-9e18-d032468a6ee6 GUO Feng, ZHEN Lei, GE Huangxu, et al. Infrared image segmentation method based on fuzzy clustering with similarity thresholding[J]. Infrared Technology, 2022, 44(8): 863-869. http://hwjs.nvir.cn/article/id/2db6a354-f9b9-4fd3-9e18-d032468a6ee6
[7] 王晓飞, 胡凡奎, 黄硕. 基于分布信息直觉模糊c均值聚类的红外图像分割算法[J]. 通信学报, 2020, 41(5): 120-129. https://www.cnki.com.cn/Article/CJFDTOTAL-TXXB202005013.htm WANG Xiaofei, HU Fankui, HUANG Shuo. Infrared image segmentation algorithm based on distribution information intuitionistic fuzzy c-means clustering[J]. Journal on Communications, 2020, 41(5): 120-129. https://www.cnki.com.cn/Article/CJFDTOTAL-TXXB202005013.htm
[8] 邢致恺, 贾鹤, 宋文龙. 基于莱维飞行樽海鞘群优化算法的多阈值图像分割[J]. 自动化学报, 2021, 47(2): 363-377. https://www.cnki.com.cn/Article/CJFDTOTAL-MOTO202102009.htm XING Zhikai, JIA He, SONG Wenlong. Levy flight trajectory-based salpswarm algorithm for multilevel thresholding image segmentation[J]. Acta Automatica Sinica, 2021, 47(2): 363-377. https://www.cnki.com.cn/Article/CJFDTOTAL-MOTO202102009.htm
[9] 李云红, 张秋铭, 周小计, 等. 基于形态学及区域合并的分水岭图像分割算法[J]. 计算机工程与应用, 2020, 56(2): 190-195. https://www.cnki.com.cn/Article/CJFDTOTAL-JSGG202002028.htm LI Yunhong, ZHANG Qiuming, ZHOU Xiaoji, et al. Watershed image segmentation algorithm based on morphology and region merging[J]. Computer Engineering and Applications, 2020, 56(2): 190-195. https://www.cnki.com.cn/Article/CJFDTOTAL-JSGG202002028.htm
[10] 周可慧, 廖志伟, 肖异瑶, 等. 基于改进CNN的电力设备红外图像分类模型构建研究[J]. 红外技术, 2019, 41(11): 1033-1038. https://www.cnki.com.cn/Article/CJFDTOTAL-HWJS201911007.htm ZHOU Kehui, LIAO Zhiwei, XIAO Yiyao, et al. Construction of infrared image classification model for power equipment based on improved CNN[J]. Infrared Technology, 2019, 41(11): 1033-1038. https://www.cnki.com.cn/Article/CJFDTOTAL-HWJS201911007.htm
[11] Krinidis S, Chatzis V. A Robust fuzzy local information c-means clustering algorithm[J]. IEEE Transactions on Image Processing, 2010, 19(5): 1328-1337. DOI: 10.1109/TIP.2010.2040763
[12] GONG M G, ZHOU Z Q, MA J J. Change detection in synthetic aperture radar images based on image fusion and fuzzy clustering[J]. IEEE Transactions on Image Processing, 2012, 21(4): 2141-2151. DOI: 10.1109/TIP.2011.2170702
[13] 王小鹏, 张永芳, 王伟, 等. 基于自适应滤波的快速广义模糊C均值聚类图像分割[J]. 模式识别与人工智能, 2018, 31(11): 1040-1046. https://www.cnki.com.cn/Article/CJFDTOTAL-MSSB201811011.htm WANG Xiaopeng, ZHANG Yongfang, WANG Wei, et al. Image segmentation using fast generalized fuzzy c-means clustering based on adaptive filtering[J]. Pattern Recognition and Artificial Intelligence, 2018, 31(11): 1040-1046. https://www.cnki.com.cn/Article/CJFDTOTAL-MSSB201811011.htm
[14] 李昌兴, 张晓璐, 雷柳. 基于粒子群优化非局部模糊聚类图像分割算法[J]. 光电子·激光, 2019, 30(7): 759-767. https://www.cnki.com.cn/Article/CJFDTOTAL-GDZJ201907013.htm LI Changxing, ZHANG Xiaolu, LEI Liu. Non-local fuzzy clustering image segmentation algorithm based on particle swarm optimization[J]. Journal of Optoelectronics Laser, 2019, 30(7): 759-767. https://www.cnki.com.cn/Article/CJFDTOTAL-GDZJ201907013.htm
[15] 火忠彩. 基于改进FCM的图像分割算法研究[D]. 兰州: 兰州交通大学, 2019. HUO Zhongcai. Research of Image Segmentation Algorithm Based on Improved FCM[D]. Lanzhou: Lanzhou Jiaotong University, 2019.
[16] WANG Qingsheng, WANG Xiaopeng, FANG Chao, et al. Robust fuzzy c-means clustering algorithm with adaptive spatial & intensity constraint and membership linking for noise image segmentation[J]. Applied Soft Computing, 2020, 92: 106318. DOI: 10.1016/j.asoc.2020.106318
[17] LEI Tao, JIA Xiaohong, ZHANG Yanning, et al. Significantly fast and robust fuzzy c-means clustering algorithm based on morphological reconstruction and membership filtering[J]. IEEE Transactions on Fuzzy Systems, 2018, 26(5): 3027-3041. DOI: 10.1109/TFUZZ.2018.2796074
[18] 刘剑秋, 阮秋琦. 形态学重建滤波器的研究与应用[J]. 通信学报, 2002(1): 116-121. https://www.cnki.com.cn/Article/CJFDTOTAL-TXXB200201016.htm LIU Jianqiu, RUAN Qiuqi. The research and application of morphological filter by reconstruction[J]. Journal on Communications, 2002(1): 116-121. https://www.cnki.com.cn/Article/CJFDTOTAL-TXXB200201016.htm
[19] 袁芊芊, 邓洪敏, 王晓航. 基于超像素快速模糊C均值聚类与支持向量机的柑橘病虫害区域分割[J]. 计算机应用, 2021, 41(2): 563-570. https://www.cnki.com.cn/Article/CJFDTOTAL-JSJY202102039.htm YUAN Qianqian, DENG Hongmin, WANG Xiaohang. Citrus disease and insect pest area segmentation based on superpixel fast fuzzy C -means clustering and support vector machine[J]. Journal of Computer Applications, 2021, 41(2): 563-570. https://www.cnki.com.cn/Article/CJFDTOTAL-JSJY202102039.htm
[20] 刘岳, 王小鹏, 王金全, 等. 基于形态学重建和梯度分层修正的分水岭脑肿瘤分割[J]. 计算机应用研究, 2015, 32(8): 2487-2491. https://www.cnki.com.cn/Article/CJFDTOTAL-JSYJ201508061.htm LIU Yue, WANG Xiaopeng, WANG Jinquan, et al. Watershed algorithm for brain tumor segmentation based on morphological reconstruction and gradient layered modification[J]. Application Research of Computers, 2015, 32(8): 2487-2491. https://www.cnki.com.cn/Article/CJFDTOTAL-JSYJ201508061.htm
[21] 王小鹏, 李新娜, 杨文婷. 自适应形态学重建的遥感图像道路提取[J]. 测绘科学, 2022, 47(5): 83-90. https://www.cnki.com.cn/Article/CJFDTOTAL-CHKD202205011.htm WANG Xiaopeng, LI Xinna, YANG Wenting. Road extraction from remote sensing image based on adaptive morphological reconstruction[J]. Science of Surveying and Mapping, 2022, 47(5): 83-90. https://www.cnki.com.cn/Article/CJFDTOTAL-CHKD202205011.htm
[22] LEI Tao, JIA Xiaohong, LIU Tongliang, et al. Adaptive morphological reconstruction for seeded image segmentation[J]. IEEE Transactions on Image Processing: a Publication of the IEEE Signal Processing Society, 2019, 28(11): 5510-5523. DOI: 10.1109/TIP.2019.2920514
[23] 陈霄, 沈洪健, 李菲, 等. 基于谱残差与笔画宽度变换的显著性文本检测方法[J]. 吉林大学学报: 理学版, 2017, 55(6): 1528-1532. https://www.cnki.com.cn/Article/CJFDTOTAL-JLDX201706033.htm CHEN Xiao, SHEN Hongjian, LI Fei, et al. Salient text detection method based on spectral residual and stroke width transform[J]. Journal of Jilin University: Science Edition, 2017, 55(6): 1528-1532. https://www.cnki.com.cn/Article/CJFDTOTAL-JLDX201706033.htm
[24] 沈洪健. 基于农用车辆牌照特征识别系统研究[D]. 长春: 吉林农业大学, 2020. SHEN Hongjian. Research on the Feature Recognition System of Agricultural Vehicle License Plates[D]. Changchun: Jilin Agricultural University, 2020.
[25] 罗其俊, 曹志芬, 牛国臣. 基于加强模糊聚类的航空行李图像超像素分割[J]. 航空学报, 2020, 41(S2): 205-212. https://www.cnki.com.cn/Article/CJFDTOTAL-HKXB2020S2023.htm LUO Qijun, CAO Zhifen, NIU Guochen. Super-pixel segmentation of air baggage image based on enhanced fuzzy clustering[J]. Acta Aeronautica et Astronautica Sinica, 2020, 41(S2): 205-212. https://www.cnki.com.cn/Article/CJFDTOTAL-HKXB2020S2023.htm
[26] 张聪炫, 裴刘继, 陈震, 等. FRFCM聚类与深度优化的RGBD场景流计算[J]. 电子学报, 2020, 48(7): 1380-1386. https://www.cnki.com.cn/Article/CJFDTOTAL-DZXU202007018.htm ZHANG Congxuan, PEI Liuji, CHEN Zhen, et al. RGBD scene flow estimation based on FRFCM clustering and depth optimization[J]. Acta Electronica Sinica, 2020, 48(7): 1380-1386. https://www.cnki.com.cn/Article/CJFDTOTAL-DZXU202007018.htm
[27] JIA Xiaohong, LEI Tao, DU Xiaogang, et al. Robust self-sparse fuzzy clustering for image segmentation[J]. IEEE Access, 2020, 9: 146182-146195. http://www.xueshufan.com/publication/3048714307
[28] LEI Tao, LIU Peng, JIA Xiaohong, et al. Automatic fuzzy clustering framework for image segmentation[J]. IEEE Transactions on Fuzzy Systems, 2020, 9(28) : 2078-2092.
[29] LI Chunming, HUANG Rui, DING Zhaohua, et al. A level set method for image segmentation in the presence of intensity inhomogeneities with application to MRI[J]. IEEE Transactions on Image Processing, 2011, 20(7): 2007-2016.
[30] 宋小鹏, 李国熊, 张权, 等. 结合局部和非局部信息的FCM图像分割算法[J]. 火力与指挥控制, 2020, 45(10): 28-33. https://www.cnki.com.cn/Article/CJFDTOTAL-HLYZ202010006.htm SONG Xiaopeng, LI Guoxiong, ZHANG Quan, et al. An FCM algorithm incorporating with local and non-local information for image segmentation[J]. Fire Control & Command Control, 2020, 45(10): 28-33. https://www.cnki.com.cn/Article/CJFDTOTAL-HLYZ202010006.htm
计量
- 文章访问数: 75
- HTML全文浏览量: 13
- PDF下载量: 17
