Nondestructive detection of radish root phenotype based on electrical resistance tomography and ResNet
-
摘要:目的
针对现有根系表型检测方法存在价格昂贵、需要专人操作以及无法对根系表型进行原位无损检测等问题，提出一种基于电阻层析成像技术(Electrical resistance tomography，ERT)和深度残差神经网络(Deep residual network，ResNet)的萝卜根系表型无损检测方法。
方法首先，利用COMSOL软件对萝卜−琼脂场域不同情况的ERT正问题进行仿真分析，并获得大量边界电压数据；然后，基于ResNet对萝卜−琼脂场域的内部电导率分布与边界电压之间的非线性映射关系建立模型，对萝卜−琼脂场域进行图像重建；最后，基于ERT研制一套萝卜根系表型检测装置，并进行试验验证。
结果基于ERT和ResNet的萝卜根系表型检测方法能够实现萝卜根系表型的可持续无损检测，试验装置操作简单、成本低，图像重建相对误差小于5%。
结论基于ERT的萝卜根系表型检测方法可以实现对萝卜根系表型的无损检测；结合ResNet算法，成像精度较高。该方法可有效应用于萝卜根系表型的检测。
Abstract:ObjectiveIn view of the problems that the existing experimental equipment for root phenotype detection is expensive, requires special personnel to operate, and cannot perform rapid in situ non-destructive detection of root phenotypes, this paper proposes a non-destructive detection method for radish root phenotype based on electrical resistance tomography (ERT) and deep residual network (ResNet).
MethodFirstly, the ERT positive problem of the radish-agar field for different cases was performed by COMSOL, and a large amount of boundary voltage data was obtained. Secondly, the nonlinear mapping relationship between the internal conductivity distribution and the boundary voltage inside the radish-agar field was modeled based on ResNet, and image of radish-agar field were reconstructed. Finally, a set of radish root phenotype detection device was developed based on ERT, and the experimental verification was carried out.
ResultThe radish root phenotype detection method based on ERT and ResNet could achieve sustainable nondestructive detection of radish root phenotype, and the experimental setup was simple to operate and low cost, while the relative error of image reconstruction was less than 5%.
ConclusionThe method of radish root phenotype detection based on ERT can achieve nondestructive detection of radish root phenotype with high imaging accuracy combining with ResNet algorithm. This method can be effectively used for the detection of radish root phenotypes.
-
-
![]()
图 1 残差单元(a)和残差网络结构(b)
Figure 1. Residual unit (a) and framework of residual network (b)
![]()
图 5 萝卜根系表型检测实物图
a：变压器；b：STM32f103zet6系统板；c：数据测量系统电路板；d：电极阵列；e：琼脂；f：萝卜
Figure 5. Image of radish root phenotype detection by ERT
a: Transformer; b: STM32f103zet6 system board; c: Circuit board of data measurement system; d: Electrode array; e: Agar; f: Radish
![]()
图 6 空场(a、d)及混合场(b、c、e、f)仿真
Figure 6. Empty field (a, d) and mixed field (b, c, e, f) simulation
![]()
图 7 单根(a、b)及双根 (c、d) 不同位置仿真图像重建
Figure 7. Simulation image reconstruction of single root (a, b) and double root (c, d) at different positions
表 1 仿真图像重建相对误差
Table 1 Relative image error (RIE) of simulated image reconstruction
仿真直径/cm Simulation diameter 单根中心 Single root center 单根非中心 Single root non-center 双根 Double root 重建直径/cm Reconstruction diameter 相对误差/% RIE 重建直径/cm Reconstruction diameter 相对误差/% RIE 重建直径/cm Reconstruction diameter 相对误差/% RIE 1 0.86 14.00 0.84 16.00 0.88 12.00 2 2.06 3.00 1.92 4.00 1.92 4.00 3 3.14 4.67 3.11 3.67 3.07 2.33 4 3.90 2.50 3.84 4.00 3.88 3.00 5 4.84 3.20 4.86 2.80 5.14 2.80 6 5.78 3.67 6.19 3.17 6.12 2.00 7 7.30 4.29 7.16 2.29 6.77 3.29 8 7.80 2.50 7.82 2.25 7.82 2.25 
下载: 导出CSV
表 2 萝卜图像重建相对误差
Table 2 Relative image error (RIE) of radish image reconstruction
层 Tier 1个萝卜 One radish 2个萝卜 Two radishes 2个萝卜 Two radishes 实测直径/cm Measured diameter 重建直径/cm Reconstruction diameter 相对误 差/% RIE 实测直径1/cm Measured diameter1 重建直径1/cm Reconstruction diameter1 相对误 差/% RIE 实测直径2/cm Measured diameter2 重建直径2/cm Reconstruction diameter2 相对误 差/% RIE 1 2.18 2.25 3.49 1.47 1.41 3.75 2.05 2.12 3.21 2 4.51 4.33 4.12 2.89 3.00 3.73 4.45 4.28 3.69 3 6.23 6.00 3.91 4.44 4.59 3.52 5.41 5.21 3.77 4 6.85 6.62 3.30 5.44 5.56 2.24 6.22 6.05 2.70 5 7.50 7.28 2.91 5.96 5.87 1.48 6.51 6.36 2.33 6 7.80 7.64 2.13 6.03 6.18 2.55 6.69 6.53 2.39 7 7.90 7.64 3.34 6.36 6.53 2.80 6.90 6.80 1.48 8 7.71 7.46 3.27 6.55 6.67 1.80 6.84 6.53 4.53
下载: 导出CSV
-
[1] 刘志芳. 白萝卜的开发价值及其栽培技术[J]. 种子科技, 2019, 37(11): 63-64. doi: 10.3969/j.issn.1005-2690.2019.11.047 [2] HE W, CHEN Y, YIN Z. Adaptive neural network control of an uncertain robot with full-state constraints[J]. IEEE Transactions on Cybernetics, 2016, 46(3): 620-629. doi: 10.1109/TCYB.2015.2411285
[3] 魏林, 梁志怀. 白萝卜肉质根开裂的主要原因及其防治方法[J]. 长江蔬菜, 2017(3): 48-49. [4] HUDEK C, STURROCK C J, ATKINSON B S, et al. Root morphology and biomechanical characteristics of high altitude alpine plant species and their potential application in soil stabilization[J]. Ecological Engineering, 2017, 109: 228-239. doi: 10.1016/j.ecoleng.2017.05.048
[5] 李燕丽, 王昌昆, 卢碧林, 等. 基于微根管技术的盐胁迫下小麦根系生长原位监测方法[J]. 土壤学报, 2021, 58(3): 599-609. [6] ZHAO X, XING L, SHEN S, et al. Non-destructive 3D geometric modeling of maize root-stubble in-situ via X-ray computed tomography[J]. International Journal of Agricultural and Biological Engineering, 2020, 13(3): 174-179. doi: 10.25165/j.ijabe.20201303.5268
[7] PFLUGFELDER D, KOCHS J, KOLLER R, et al. The root system architecture of wheat establishing in soil is associated with varying elongation rates of seminal roots: Quantification using 4D magnetic resonance imaging[J]. Journal of Experimental Botany, 2022, 73(7): 2050-2060. doi: 10.1093/jxb/erab551
[8] PFLUGFELDER D, METZNER R, VAN DUSSCHOTEN D, et al. Non-invasive imaging of plant roots in different soils using magnetic resonance imaging (MRI)[J]. Plant Methods, 2017, 13: 102. doi: 10.1186/s13007-017-0252-9.
[9] 徐汇, 徐元元, 李光辉. 基于探地雷达的果树根系检测试验与分析[J]. 江苏农业科学, 2022, 50(2): 170-177. doi: 10.15889/j.issn.1002-1302.2022.02.029 [10] 颜华, 胡丽娟, 王伊凡, 等. 基于改进灵敏度矩阵的ERT图像重建[J]. 仪器仪表学报, 2018, 39(5): 241-248. doi: 10.19650/j.cnki.cjsi.J1703034 [11] LI F, TAN C, DONG F. Electrical resistance tomography image reconstruction with densely connected convolutional neural network[J]. IEEE Transactions on Instrumentation and Measurement, 2021, 70: 4500811. doi: 10.1109/TIM.2020.3013056.
[12] SONG Z, WEN J, WAN N, et al. An optimization algorithm H-GVSPM for electrical impedance tomography[J]. IEEE Sensors Journal, 2022: 1-8. doi: 10.1109/JSEN.2022.3170838.
[13] 王兴龙, 王立海. 白桦模拟缺陷材的电阻层析成像规律探究[J]. 林业工程学报, 2016, 1(2): 106-111. [14] 徐梓敬, 贾培, 吴楠, 等. GA-SVM在木材缺陷识别中的应用[J]. 传感器与微系统, 2019, 38(9): 153-156. doi: 10.13873/J.1000-9787(2019)09-0153-04 [15] 段国秀, 贾小旭, 白晓, 等. 电阻率层析成像法在土壤水文学中的应用: 基于CiteSpace的计量分析[J]. 土壤通报, 2021, 52(6): 1447-1459. [16] 盛丰, 文鼎, 熊祎玮, 等. 基于电阻率层析成像技术的农田土壤优先流原位动态监测[J]. 农业工程学报, 2021, 37(8): 117-124. doi: 10.11975/j.issn.1002-6819.2021.08.013 [17] WEIGAND M, KEMNA A. Multi-frequency electrical impedance tomography as a non-invasive tool to characterize and monitor crop root systems[J]. Biogeosciences, 2017, 14(4): 921-939. doi: 10.5194/bg-14-921-2017
[18] CORONA D D J, SOMMER S, ROLFE S A, et al. Electrical impedance tomography as a tool for phenotyping plant roots[J]. Plant Methods, 2019, 15: 49. doi: 10.1186/s13007-019-0438-4.
[19] 李峰, 谭超, 董峰. 全连接深度网络的电学层析成像算法[J]. 工程热物理学报, 2019, 40(7): 1526-1531. [20] 许辰航, 陈继明, 刘伟楠, 等. 基于深度残差网络的GIS局部放电PRPD谱图模式识别[J]. 高电压技术, 2022, 48(3): 1113-1123. doi: 10.13336/j.1003-6520.hve.20201663 [21] LI B, WANG J M, WANG Q, et al. An improved generalized finite element method for electrical resistance tomography forward model[J]. Journal of Electrical Engineering & Technology, 2019, 14(6): 2595-2606.
[22] 仝卫国, 曾世超, 张立峰, 等. 基于深度残差神经网络的电阻层析成像及流型辨识方法[J]. 系统仿真学报, 2022, 34(9): 2028-2036. doi: 10.16182/j.issn1004731x.joss.21-0294
