Multi-frequency complex conductivity sparse imaging of plant root zone based on space-frequency correlation

The root zone provides indispensable water and nutrients for plants and monitoring the dynamics of the root zone is highly important. However, due to the opacity and complexity of this zone, in situ characterization of the root zone poses many challenges. A sparse Bayesian learning complex model based on spatial frequency corre-lation (SF-SBLC) was proposed for fast nondestructive imaging of the growth trends in the root zone of plants, and multi-frequency electrical impedance tomography (mfEIT) was used to reconstruct the complex conductivity distribution of the root zone. To achieve the best sparse coefficients and enhance the quality of the reconstructed images, the SF-SBLC learned and utilized the spatial and frequency correlations of the complex conductivities at different frequencies. The reconstruction algorithm was further optimized using a concave-convex process method and a multi-measure vector model, which decreased its time complexity. When compared to conven-tional mfEIT reconstruction algorithms, the experimental findings demonstrated that the SF-SBLC could greatly reduce imaging artifacts, generate higher reconstruction performance (in terms of the image correlation coef-ficient, relative image error, resolution, position error, shape deformation, and ringing), and shorten imaging time. Moreover, the SF-SBLC was able to recreate the root zone's shape, size, and location in both a nutrient solution and a horticulture substrate. Finally, the effective volume of the root zone is defined and calculated based on the 3D EIT image. This technique offers a prospective tool that might be useful for monitoring the root growth trends and expansion rates of potted plants.

Automated summary

A sparse Bayesian learning complex model based on spatial frequency correlation and multi-frequency electrical impedance tomography was proposed for fast nondestructive imaging of the growth trends in the root zone of plants. The technique could reduce imaging artifacts, generate higher reconstruction performance, and recreate the root zone's shape, size, and location.