Flotation condition recognition based on multi-scale convolutional neural network and LBP algorithm
-
摘要: 针对泡沫浮选加药状态检测困难、识别效率低和主观性强等问题，提出了一种结合多尺度卷积神经网络(CNN)特征及改进局部二值模式(LBP)计算方法的核随机权神经网络(K-RVFLNs)浮选工况识别方法。首先，对泡沫浮选图像进行非下采样Shearlet多尺度分解，将原始图像分解为不同频率尺度，设计多通道CNN网络对多尺度图像进行特征提取; 再通过改进LBP算法提取特征作为补充，将CNN提取的图像特征与LBP特征进行融合; 最后，通过核随机权神经网络映射到更高维空间进行分类决策，实现浮选加药状态的精确识别。实验结果表明，采用多尺度CNN及LBP-TOP特征融合的方法识别的精度比传统LBP算法提高了5.34 %，比采用单CNN特征的方法提高了3.76 %，结合K-RVFLNs实现浮选工况分类准确率高达96.38 %，识别精度和稳定性较现有方法有较大提升，且减少了人工干预，有利于提高生产效率。
-
关键词:
- 图像处理 /
- 卷积神经网络 /
- 非下采样Shearlet变换 /
- 局部二值模式 /
- 随机权神经网络
Abstract: Aiming at the problems of difficult detection, low recognition efficiency and strong subjectivity of foam flotation dosing state, a flotation condition recognition method of nuclear random-weight neural networks(K-RVFLNs)combining multi-scale CNN characteristics and improved local binary patterns(LBP)calculation methods is proposed.Firstly, non-downsampling Shearlet multi-scale decomposition(NSST)is performed on the bubble flotation image, the original image is decomposed into different frequency scales, and a multi-channel CNN network is designed to extract features from the multi-scale image.By improving the LBP algorithm to extract features as a supplement, the image features extracted by CNN are fused with LBP features; Finally, the classification decision is made by mapping to a higher dimensional space by the nuclear stochastic right neural network, and the accurate identification of flotation dosing state is realized.Experimental results show that the method of multi-scale CNN and LBP-TOP feature fusion is 5.34 % higher than that of the traditional LBP algorithm, 3.76 % higher than that of single CNN feature recognition method, and the accuracy of flotation condition classification is as high as 96.38 % in combination with K-RVFLNs, and the recognition accuracy and stability are greatly improved compared with the existing methods, and this method reduces manual intervention and is conducive to improving production efficiency. -
图 8 深度双隐层自编码核随机权神经网络
Figure 8. Deep double hidden layer self-encoded kernel random-weight neural network
图 9 结合CNN和LBP特征的K-RVFLNs工况识别模型
Figure 9. K-RVFLNs condition recognition model combining CNN and LBP features
表 1 不同方法的识别效果
Table 1. Recognition effect of different methods
类别 特征提取方法 分类算法 运行时间/s 准确率/% 方法1 气泡大小和形状的联合分布 最小二乘分类 1.053 85.62 方法2 泡沫大小分布、灰度特征、纹理特征 支持向量机 1.851 84.61 方法3 泡沫纹理特征 自适应权重支持向量机 2.158 89.78 方法4 气 K均值聚类 3.614 93.21 方法5 双模态特征提取 随机权神经网络 2.892 95.98 方法6 Alexnet网络迁移学习 随机森林 2.023 93.41 方法7 轻量型卷积神经网络模块 多层感知器 2.821 93.56 本文方法 NSST-CNN特征+LBP-TOP特征 核随机权神经网络 3.028 96.38 
下载: 导出CSV
-
[1] Zhang J, Tang Z H, Ai M X, et al. Nonlinear modeling of the relationship between reagent dosage and flotation froth surface image by Hammerstein-Wiener model[J]. Minerals Engineering, 2018, 120: 19-28. doi: 10.1016/j.mineng.2018.01.018 [2] 黄凌霄, 廖一鹏. 浮选气泡NSCT域多尺度等效形态特征提取及识别[J]. 光学精密工程, 2020, 28(3): 704-716. https://www.cnki.com.cn/Article/CJFDTOTAL-GXJM202003021.htmHuang Lingxiao, Liao Yipeng. Recognition and multiscale equivalent morphological features extraction of flotation bubbles in NSCT domain[J]. Optics and Precision Engineering, 2020, 28(3): 704-716. https://www.cnki.com.cn/Article/CJFDTOTAL-GXJM202003021.htm [3] 姚群力, 胡显, 雷宏. 基于多尺度卷积神经网络的遥感目标检测研究[J]. 光学学报, 2019, 39(11): 346-353. https://www.cnki.com.cn/Article/CJFDTOTAL-GXXB201911042.htmYao Qunli, Hu Xian, Lei Hong. Object detection in remote sensing images using multiscale convolutional neural networks[J]. Acta Optica Sinica, 2019, 39(11): 346-353. https://www.cnki.com.cn/Article/CJFDTOTAL-GXXB201911042.htm [4] Fu Y H, Aldrich C. Froth image analysis by use of transfer learning and convolutional neural networks[J]. Minerals Engineering, 2018, 115: 68-78. doi: 10.1016/j.mineng.2017.10.005 [5] Fu Y, Aldrich C. Flotation froth image recognition with convolutional neural networks[J]. Minerals Engineering, 2019, 132: 183-190. doi: 10.1016/j.mineng.2018.12.011 [6] Morar S H, Bradshaw D J, Harris M C. The use of the froth surface lamellae burst rate as a flotation froth stability measurement[J]. Minerals Engineering, 2012, 36/37/38: 152-159. [7] Wang X L, Chen S, Yang C H, et al. Process working condition recognition based on the fusion of morphological and pixel set features of froth for froth flotation[J]. Minerals Engineering, 2019, 128: 17-26. [8] 廖一鹏, 杨洁洁, 王志刚, 等. 基于双模态卷积神经网络自适应迁移学习的浮选工况识别[J]. 光子学报, 2020, 49(10): 173-184. https://www.cnki.com.cn/Article/CJFDTOTAL-GZXB202010020.htmLiao Yipeng, Yang Jiejie, Wang Zhigang, et al. Flotation performance recognition based on dual-modality convolutional neural network adaptive transfer learning[J]. Acta Photonica Sinica, 2020, 49(10): 173-184. https://www.cnki.com.cn/Article/CJFDTOTAL-GZXB202010020.htm [9] Li Z M, Gui W H, Zhu J Y. Fault detection in flotation processes based on deep learning and support vector machine[J]. Journal of Central South University, 2019, 26(9): 2504-2515. doi: 10.1007/s11771-019-4190-8 [10] 陈奕霏, 蔡耀仪, 李诗文. 基于轻量型卷积视觉Transformer的锑浮选工况识别[J]. 激光与电子学科进展, 2023, 60(6): 0615002.Chen Yifei, Cai Yaoyi, Li Shiwen, et al. Working condition recongnition using lightweight convolution vision transformer network for antimony flotation process[J]. Laser & Optoelectronics Progress, 2023, 60(6): 0615002. [11] Labate D, Lim W Q, Kutyniok G, et al. Sparse multidimensional representation using shearlets[C]//Optics and Photonics 2005. Proc SPIE 5914, Wavelets XI, San Diego, California, USA. 2005, 5914: 254-262. [12] Easley G, Labate D, Lim W Q. Sparse directional image representations using the discrete shearlet transform[J]. Applied and Computational Harmonic Analysis, 2008, 25(1): 25-46. doi: 10.1016/j.acha.2007.09.003 [13] Liu X B, Mei W B, Du H Q, et al. A novel image fusion algorithm based on nonsubsampled shearlet transform and morphological component analysis[J]. Signal, Image and Video Processing, 2016, 10(5): 959-966. [14] Shahdoosti H R, Khayat O. Image denoising using sparse representation classification and non-subsampled shearlet transform[J]. Signal, Image and Video Processing, 2016, 10(6): 1081-1087. [15] Wu J A, Guo R, Liu R Z, et al. Convolutional neural network target recognition for missile-borne linear array LiDAR[J]. Acta Photonica Sinica, 2019, 48(7): 701002. [16] Liu S, Qi L, Qin H F, et al. Path aggregation network for instance segmentation[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City, UT, USA: IEEE, 2018: 8759-8768. [17] Rezatofighi H, Tsoi N, Gwak J, et al. Generalized intersection over union: a metric and a loss for bounding box regression[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR). Long Beach, CA, USA: IEEE, 2019: 658-666. [18] Zheng Z H, Wang P, Liu W, et al. Distance-IoU loss: faster and better learning for bounding box regression[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34(7): 12993-13000. [19] 余益团. 基于LBP-top特征的新生儿疼痛表情识别研究[D]. 南京: 南京邮电大学, 2016. [20] Zhao G Y, Pietikäinen M. Dynamic texture recognition using local binary patterns with an application to facial expressions[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2007, 29(6): 915-928. [21] 赵突. 基于视频的人脸微表情识别关键技术研究[D]. 南京: 东南大学, 2018. [22] 李强. 基于视频的微表情检测与识别技术研究[D]. 南京: 东南大学, 2017. [23] 郭承玉. 人脸自发微表情识别方法研究[D]. 长沙: 国防科技大学, 2019. [24] 余益团. 基于LBP-top特征的新生儿疼痛表情识别研究[D]. 南京: 南京邮电大学, 2016. [25] Pao Y H, Takefuji Y. Functional-link net computing: theory, system architecture, and functionalities[J]. Computer, 1992, 25(5): 76-79. [26] Huang G B. An insight into extreme learning machines: random neurons, random features and kernels[J]. Cognitive Computation, 2014, 6(3): 376-390. [27] Sinha A, Soun T, Deb K. Using Karush-Kuhn-Tucker proximity measure for solving bilevel optimization problems[J]. Swarm and Evolutionary Computation, 2019, 44: 496-510. [28] Kim J, Lee B. Identification of Alzheimer's disease and mild cognitive impairment using multimodal sparse hierarchical extreme learning machine[J]. Human Brain Mapping, 2018, 39(9): 3728-3741. [29] 廖一鹏, 张进, 王志刚, 等. 结合双模多尺度CNN特征及自适应深度KELM的浮选工况识别[J]. 光学精密工程, 2020, 28(8): 1785-1798. https://www.cnki.com.cn/Article/CJFDTOTAL-GXJM202008018.htmLiao Yipeng, Zhang Jin, Wang Zhigang, et al. Flotation performance recognition based on dual-modality multiscale CNN features and adaptive deep learning KELM[J]. Optics and Precision Engineering, 2020, 28(8): 1785-1798. https://www.cnki.com.cn/Article/CJFDTOTAL-GXJM202008018.htm -
点击查看大图
图(15) / 表(1)
计量
- 文章访问数: 261
- HTML全文浏览量: 107
- PDF下载量: 38
- 被引次数: 0