基于电阻抗成像的植物单根断层图像重建

为了实现植物根系结构形态的原位检测，植物根系单根的断层图像重建研究非常必要。该文应用动态电阻抗成像技术对土壤-树根模拟系统进行了图像重建，对重建图像进行二值化处理并求出树根在该系统中的位置坐标、面积和形状，并以树根的位置坐标、面积及形状的相对偏差为指标，分析了图像重建算法、土壤含水率2个因素对成像质量的影响。结果表明：电阻抗成像技术能够实现树木单根断层图像重建；将重建图像进行二值化处理，所得图像更加直观，便于定量评价成像质量；基于Newton单步残差正则化的一步牛顿误差重构（Newton's one-step error reconsruction，NOSER）算法，比基于全变差正则化的主双-内点模式（primal dual-interior point method，PD-IPM）算法的成像质量好；土壤含水率越高成像质量越好。该研究为基于电阻抗成像技术的植物根系结构形态的图像重建提供参考。

Tomographic image reconstruction of plant single root by electrical impedance tomography

Abstract: Root architecture plays an important role in plant nutrient and moisture absorption, so the study on plant root architecture and growth pattern is crucial to improve ecological environment and plant productivity. It is also necessary to obtain root architecture of plant cheaply, fastly and easily. Electrical impedance tomography (EIT) was used to produce images according to electrical parameters distribution within the studied region. The electrical parameters were calculated using the resulted potentials on the electrodes by which sinusoidal electrical current was applied into the imaging domain. This method has been extensively developed for both medical and industrial use during the past decades as visualization and measurement technique, but rarely used in pant root architecture reconstruction. EIT has several advantages over other tomography techniques such as X-ray computed tomography (X-CT), magnetic resonance imaging (MRI) and ground penetrating radar (GPR) due to its portability, safety and low cost. To verify whether EIT is applicable for reconstruction of plant root segment and explore the effect of image reconstruction algorithm and soil water content on image reconstruction quality, an experimental system composed of LCR analyzer, PC and self-made experimental cell with 16 electrodes was built and adjacent measurement mode was used to obtain the voltage data for imaging. These voltage data at different soil water contents was measured, Newton's One-Step Error Reconstructor (NOSER) and the Primal Dual-Interior Point Method algorithm (PD-IPM) were used to obtain the reconstruction images. During the procedure of reconstruction, the reconstruction image domain was divided into 3922 units by triangular elements. The reconstruction algorithms were mainly used to calculate the electrical resistivity of the corresponding triangular unit. Then the colors representing different values mapped out the reconstruction images. On account of the apparent color differences between the root and soil, the reconstruction of the root was realized. The colorful reconstruction image was firstly processed into binarized image, then the location, area and circularity of root were calculated by pixel accumulation method. Coincidence degree index between the calculated and actual value was used to evaluate the reconstruction image quality. The results showed that the tomographic image reconstruction of single root for plant can be achieved by the EIT method effectively. The binarized image was more intuitive and can be used easily to evaluate the quality of reconstruction. NOSER and PD-IPM both reflected the shape of the root well, but the coordinate of root centroid was not presented accurately, the former can get more accurate information on the size of the roots. Soil water content had a great effect on the area and location of the root in the reconstructed image, higher moisture content produced more accurate reconstruction image, but it has little effect on the root shape in the reconstructed image. The study provides a reference for the further application of electrical impedance tomography technique on nondestructive and in situ detection of roots, which will provide a new idea for the reconstruction of root architecture.