针对传统图像处理算法对重叠颗粒的分割困难,引入Mask R-CNN深度学习算法并做针对性改进,通过调整残差网络ResNet-101加速训练,提出双FPN结构实现全局特征融合,使用Soft-NMS避免重叠颗粒漏检。设计了颗粒重叠图像实验系统,采集单一球形、球形与不规则混合多分散颗粒重叠图像用于分割研究。实验结果表明:该方法对混合颗粒分类准确率为91%,召回率为92%,均优于传统算法;其应用于含气泡的一水柠檬酸结晶过程中结晶的在线测量,所得结晶颗粒中位径误差为3.8%,数目误差为-1.3%。所提方法为混合颗粒的重叠图像分析提供了思路,后续有望解决图像法结晶过程后期在线监测乏力与气泡干扰的问题。
Abstract
Mask R-CNN was introduced to overcome the segmentation difficulties of traditional image processing algorithms for overlapped particle images.By adjusting residual network ResNet-101 to accelerate training, a double FPN structure was proposed to achieve global feature fusion, and soft-NMS was used to avoid o overlapped particle missing detection.A particle overlapped image experiment system was designed to acquire single spherical, spherical and irregular mixed multi-dispersed overlapped particle images for segmentation analysis.The experimental results show that the present classification accuracy is 91%, and the recall rate is 92%, which are both better than the traditional algorithms.When applied to the real-time measurement of crystallization and bubbles in the crystallization process of citric acid monohydrate, the method yields the errors around 3.8% for median diameter and -1.6% for the counting number of crystal particles.The proposed method provides a clue for analysis of overlapped mixed particle images, which is expected to solve the problems of image analysis at late stage of the crystallization process and eliminate the interference of bubbles involved during real-time monitoring.
关键词
计量学;混合颗粒;重叠图像;颗粒分割;粒径分布;一水柠檬酸 /
结晶;深度学习
Key words
metrology /
mixed particle /
overlapped image /
particle segmentation /
size distribution /
citric acid monohydrate /
crystallization /
deep learning
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
［1］Pei Y W, Hinchliffe B A, Minelli C.Measurement of the Size Distribution of Multimodal Colloidal Systems by Laser Diffraction［J］.ACS omega,2021,6 (22):14049-14058.
［2］曲佩玙,蒋瑜,苏明旭,等.基于图像光散射颗粒粒度测量方法研究［J］. 计量学报,2021, 42 (4): 469-475.
Qu P Y,Jiang Y,Su M X,et al.Research on Measurement Method of Particle Size Based on Image Light Scattering［J］. Acta Metrologica Sinica.2021,42(4):469-475.
［3］吴丹,张国城,赵晓宁.光散射法颗粒物监测仪粒径识别检测装置的搭建及方法研究［J］. 计量学报,2021, 42 (3): 388-394.
Wu D, Zhang G C, Zhao X N.Research on Construction and Method of Particle Size Recognition and Detection Device for Light Scattering Particles Monitor［J］. Acta Metrologica Sinica, 2021, 42 (3): 388-394.
［4］贾楠,顾建飞,苏明旭.基于超声谱分析的颗粒粒度测量研究［J］. 计量学报,2019, 40 (3): 116-121.
Jia N, Gu J F, Su M S.Characterization of Particle Size Distribution Based on Ultrasonic Spectra Analysis［J］. Acta Metrologica Sinica, 2019, 40 (3): 116-121.
［5］周建华,房怀英,杨建红,等.图像法集料粒径检测表征参数的选择及实验研究［J］. 计量学报,2018, 39 (6): 783-790.
Zhou J H, Fang H Y, Yang J H, et al.Selection of characterization parameters for image aggregate particle size detection and experimental research［J］. Acta Metrologica Sinica, 2018,39 (6): 783-790.
［6］张莉,储茂祥,邓鑫,等.基于图的矿石颗粒图像检测技术［J］. 合肥工业大学学报 (自然科学版),2021, 44 (9): 1193-1197.
Zhang L, Chu M X, Deng X, et al.Graph based detection technology for mineral grains image［J］. Journal of Hefei University of Technology (Natural Science), 2021, 44 (9): 1193-1197.
［7］潘志成,赵陆海波,张彪,等.多尺度气泡尺寸分布数字图像测量方法研究［J］. 仪器仪表学报,2019, 40 (7): 129-137.
Pan Z C, Zhao L H B, Zhang B, et al.Multi scale bubble size distribution measurement by digital imaging technique［J］. Chinese Journal of Scientific Instrument, 2019, 40 (7): 129-137.
［8］Miao H S,Xiao C Y.Simultaneous Segmentation of Leukocyte and Erythrocyte in Microscopic Images Using a Marker-Controlled Watershed Algorithm［J］.Computational and mathematical methods in medicine,2018(9):1-9.
［9］蔡改贫,刘占,汪龙,等.基于形态学优化处理的标记符分水岭矿石图像分割［J］. 科学技术与工程,2020, 20 (23): 9497-9502.
Cai G P, Liu Z, Wang L, et al.Marker Watershed Ore Image Segmentation Based on Morphological Optimization Processing［J］. Science Technology and Engineering, 2020, 20 (23): 9497-9502.
［10］郭观凯,刘伟,余玲玲,等.基于改进FAST与分水岭算法的颗粒图像分割［J］. 中国粉体技术,2019, 25 (2): 61-67.
Guo G K, Liu W, Yu L L, et al.Image segmentation of touching particles based on improved FAST and watershed algorithm［J］. China Powder Science and Technology, 2019, 25 (2): 61-67.
［11］Zafari S, Eerola T, Sampo J, et al.Segmentation of Partially Overlapping Nanoparticles Using Concave Points ［C］//11th International Symposium on Advances in Visual Computing.2015.
［12］Long J, Shelhamer E, Darrell T.Fully convolutional networks for semantic segmentation［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 39 (4): 640-651.
［13］Badrinarayanan V, Kendall A, Cipolla R. SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39 (12): 2481-2495.
［14］Ronneberger O, Fischer P, Brox T.U-Net: Convolutional Networks for Biomedical Image Segmentation［C］//Proceedings of International Conference on Medical Image Computing and Computer Assisted Intervention. Berlin, Germany, 2015.
［15］He K M, Gkioxari G, Dollár P, et al. Mask R-CNN［C］//Proc of IEEE International Conference on Computer Vision.2017.
［16］He K M, Zhang X, Ren S, et al.Deep residual learning for image recognition［C］//Proc of IEEE Conference on Computer Vision.Piscataway, NJ, USA, 2016.
［17］Lin T Y, Dollár P, Girshick R, et al.Feature pyramid networks for object detection［C］//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).Piscataway, NJ, USA, 2017.
［18］Bodla N, Singh B, Chellappa R, et al. Soft-NMS improving object detection with one line of code［C］//Proc of IEEE Conference on Computer Vision. Piscataway, NJ, USA,2017.
［19］贾楠,田昌,苏明旭.无水醋酸钠结晶过程中析晶温度和颗粒粒径在线测量［J］. 化工学报,2019, 70 (12): 4664-4672.
Jia N, Tian C, Su M X.In situ measurement of crystallization temperature and particle size distribution during crystallization of sodium acetate［J］. Journal of Chemical Industry and Engineering, 2019, 70 (12): 4664-4672.
［20］许潇,张文阁,池顺鑫.基于圆形拟合识别算法的颗粒粒度粒形自动测量方法研究［J］.计量科学与技术,2021,65(3):15-18.
Xu X, Zhang W G, Chi S X. Automatic Particle Measurement Method Based on Circle Fitting Aided Recognition Algorithm［J］. Metrology Science and Technology,2021,65(3):15-18.
基金
国家自然科学基金(51776129);中国航发四川燃气涡轮研究院外委课题
PDF(5745 KB)
