Epidermal Cell Image Recognition Research of Improved EfficientNet

doi:10.3778/j.issn.1002-8331.2106-0186

Abstract

Abstract: The amount of microscopic image data of traditional Chinese medicinal materials powder is small, and there are certain differences in features and shapes of different production areas and different collection environments. The traditional image classification methods have poor recognition results under cross-database condition. To solve the above problems, it proposes an improved method of deep convolutional neural network based on multi-channel fusion and SPP structure. First, it combines feature maps obtained by Canny edge detection and local binary pattern with original image to form a five-channel image and then it is sent to the network, in order to expand the data width of the network input; second, it embeds the improved SPP module in the EfficientNet network, in order to increase the depth of the network. The above methods can make the network pay more attention to the deep texture information of image, so that it is not affected by the collection environment, etc. , and solves the problem of cross-database recognition. The experimental results show that for two different batches of microscopic images of the epidermal cells of Chinese medicinal materials of 26 kinds, using library 1 as the training set and library 2 as the test set, the accuracy rate has increased by 2.7 percentage points to 81.5%, which proves the proposed research method has certain advantages for the task of classification of microscopic images of traditional Chinese medicinal materials under cross-database condition.

Key words: convolutional neural network, image classification of microscopic features of Chinese medicinal materials powder, multi-channel fusion, cross-database, deep learning

摘要： 中药材粉末显微图像数据量少，而且在不同产地、不同采集环境下的特征形态具有一定的差异，传统的图像分类方法跨库识别效果不佳。针对以上问题提出一种基于多通道融合和SPP结构的深度卷积神经网络改进方法。采用Canny边缘检测和局部二值模式的特征图与原图合并形成五通道的图像送入网络，扩充网络输入端的数据宽度；将改进的SPP模块嵌入到EfficientNet网络中，增加网络的深度。以上方法可以使网络更加注重图像的深层纹理信息，从而不受采集环境等的影响，很好地解决跨库识别问题。实验结果表明，对26种中药材粉末表皮细胞显微图像的两个不同批次数据，采用1库作为训练集，2库作为测试集，其准确率提升了2.7个百分点，达到81.5%，证明了所提研究方法对于跨库中药材显微图像分类任务具有一定的优势。

关键词: 卷积神经网络, 中药材粉末显微特征图像识别, 多通道融合, 跨数据库, 深度学习

WANG Yiding, YAO Yi, LI Yaoli, CAI Shaoqing, YUAN Yuan. Epidermal Cell Image Recognition Research of Improved EfficientNet[J]. Computer Engineering and Applications, 2022, 58(11): 200-208.

王一丁, 姚毅, 李耀利, 蔡少青, 袁媛. 改进EfficientNet的表皮细胞图像识别研究[J]. 计算机工程与应用, 2022, 58(11): 200-208.

References

[1] 王亚杰，徐心和.中药材粉末显微图像模式识别方法[J].农机化研究，2006（12）：180-183.
WANG Y J，XU X H.Pattern recognition method on micro-images of powdered Chinese medical herbs[J].Journal of Agricultural Mechanization Research，2006（12）：180-183.
[2] 李辉，韩墨，朱飞叶，等.基于扫描电镜法与X-衍射指纹图谱法浙产珍珠粉的鉴别研究[J].中国新药杂志，2013，22（23）：2817-2821.
LI H，HAN M，ZHU F Y，et al.Identification of pearl powder from Zhejiang by scanning electron microscope and X-ray diffraction analysis[J].Chinese Journal of New Drugs，2013，22（23）：2817-2821.
[3] 王耐，卢文彪，凌秀华，等.牛膝和川牛膝药材的特征提取与图像识别[J].中国药房，2017，28（12）：1670-1673.
WANG N，LU W B，LING X H，et al.Feature extraction and image recognition of achyranthes bidentata and cyathula officinalis[J].China Pharmacy，2017，28（12）：1670-1673.
[4] 王一丁，石铎，李耀利，等.基于SqueezeNet深度网络的中药材粉末显微特征图像识别研究[J].电子显微学报，2019，38（2）：130-138.
WANG Y D，SHI D，LI Y L，et al.Studies on identification of microscopic images of chinese Chinese medicinal materials powder based on SqueezeNet deep network[J].Journal of Chinese Electron Microscopy Society，2019，38（2）：130-138.
[5] 王一丁，郝晨宇，李耀利，等.基于深度学习的小样本中药材粉末显微图像识别[J].计算机应用，2020，40（5）：1301-1308.
WANG Y D，HE C Y，LI Y L，et al.Microscopic image identification for small-sample Chinese medicinal materials powder based on deep learning[J].Journal of Computer Applications，2020，40（5）：1301-1308.
[6] LECUN Y，BOSER B，DENKER J S，et al.Backpropagation applied to handwritten zip code recognition[J].Neural Computation，1989，1（4）：541-551.
[7] KRIZHEVSKY A，SUTSKEVER I，HINTON G E.Imagenet classification with deep convolutional neural networks[C]//Advances in Neural Information Processing Systems，2012：1097-1105.
[8] HE K，ZHANG X，REN S，et al.Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，Las Vegas，Jun 27-Jun 30，2016.New York：IEEE，2016：770-778.
[9] HOWARD A G，ZHU M，CHEN B，et al.Mobilenets：efficient convolutional neural networks for mobile vision applications[J].arXiv：1704.04861，2017.
[10] TAN M，LE Q.Efficientnet：rethinking model scaling for convolutional neural networks[C]//Proceedings of the International Conference on Machine Learning，Long Beach，Jun 9-Jun 15，2019：6105-6114.
[11] HE K，ZHANG X，REN S，et al.Spatial pyramid pooling in deep convolutional networks for visual recognition[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2015，37（9）：1904-1916.
[12] BOCHKOVSKIY A，WANG C Y，LIAO H Y M.Yolov4：optimal speed and accuracy of object detection[J].arXiv：2004.10934，2020.
[13] 宝音图，刘伟，牛朝阳，等.联合集成学习与EfficientNet的光学遥感图像场景分类[J].计算机工程，2021，47（10）：226-235.
BAO Y T，LIU W，NIU C Y，et al.Scene classification of optical remote sensing image based on ensemble learning and EfficientNet[J].Computer Engineering，2021，47（10）：226-235.
[14] 宋瑶，刘俊.改进U-Net的新冠肺炎图像分割方法[J].计算机工程与应用，2021，57（19）：243-251.
SONG Y，LIU J.Improved U-Net network for COVID-19 image segmentation[J].Computer Engineering and Applications，2021，57（19）：243-251.
[15] 叶紫璇，肖满生，肖哲.基于EfficientNet模型的多特征融合肺癌病理图像分型[J].湖南工业大学学报，2021，35（2）：51-57.
YE Z X，XIAO M S，XIAO Z.Lung cancer pathological image classification based on an EfficientNet model with multi-feature fusion[J].Journal of Hunan University of Technology，2021，35（2）：51-57.
[16] 于金扣，余南南，于贺，等.基于多尺度层级化注意力模型的脑血肿分割算法[J].航天医学与医学工程，2021，34（1）：44-51.
YU J K，YU N N，YU H，et al.Segmentation algorithm of intracranial hematoma based on multi-scale hierarchical attention model[J].Space Medicine & Medical Engineering，2021，34（1）：44-51.
[17] BHUMA C M，KONGARA R.Childhood medulloblastoma classification using efficientNets[C]//Proceedings of the 2020 IEEE Bombay Section Signature Conference（IBSSC），Mumbai，Dec 4-Dec 6，2020：64-68.
[18] ALHICHRI H，ALSWAYED A S，BAZI Y，et al.Classification of remote sensing images using EfficientNet-B3 CNN model with attention[J].IEEE Access，2021，9：14078-14094.
[19] 徐辉，李海翔，唐世轩，等.基于级联特征分类器的行人检测算法[J].实验室研究与探索，2021，40（2）：127-132.
XU H，LI H X，TANG S X，et al.Pedestrian detection algorithm based on cascaded feature classifier[J].Research and Exploration in Laboratory，2021，40（2）：127-132.
[20] 陈杨，石晶，刘丛浩.基于改进霍夫变换的车道线识别算法[J].汽车实用技术，2021，46（6）：42-44.
CHEN Y，SHI J，LIU C H.Lane recognition algorithm based on improved Hough transform[J].Automobile Applied Technology，2021，46（6）：42-44.
[21] HEIKKIL? M，PIETIK?INEN M，SCHMID C.Description of interest regions with local binary patterns[J].Pattern Recognition，2009，42（3）：425-436.
[22] 汤旻安，王晨雨.基于改进Canny算法与Hough变换的轨道图像边缘提取[J].铁道标准设计，2021，65（8）：60-64.
TANG M A，WANG C Y.Extraction of the edge of track images based on improved Canny algorithm and Hough transform[J].Railway Standard Design，2021，65（8）：60-64.