Multi-Image Super Resolution Framework for Detection and Analysis of Plant Roots

TL;DR

This work proposes a novel underground imaging system that captures multiple overlapping views of plant roots and integrates a deep learning-based Multi-Image Super Resolution (MISR) framework designed to enhance root visibility and detail, thereby enabling enhanced phenotypic analysis of root systems.

Abstract

Understanding plant root systems is critical for advancing research in soil-plant interactions, nutrient uptake, and overall plant health. However, accurate imaging of roots in subterranean environments remains a persistent challenge due to adverse conditions such as occlusion, varying soil moisture, and inherently low contrast, which limit the effectiveness of conventional vision-based approaches. In this work, we propose a novel underground imaging system that captures multiple overlapping views of plant roots and integrates a deep learning-based Multi-Image Super Resolution (MISR) framework designed to enhance root visibility and detail. To train and evaluate our approach, we construct a synthetic dataset that simulates realistic underground imaging scenarios, incorporating key environmental factors that affect image quality. Our proposed MISR algorithm leverages spatial redundancy across views to reconstruct high-resolution images with improved structural fidelity and visual clarity. Quantitative evaluations show that our approach outperforms state-of-the-art super resolution baselines, achieving a 2.3 percent reduction in BRISQUE, indicating improved image quality with the same CLIP-IQA score, thereby enabling enhanced phenotypic analysis of root systems. This, in turn, facilitates accurate estimation of critical root traits, including root hair count and root hair density. The proposed framework presents a promising direction for robust automatic underground plant root imaging and trait quantification for agricultural and ecological research.

Figures (6)

• Figure 1: RootCam deployed in an agriculture field. Note that there are three visible RootCam setups each highlighted with a green rectangle.
• Figure 2: A sample of root image showing root hairs captured by the RootCam.
• Figure 3: An image generated by our synthetic root generation algorithm.
• Figure 4: Architecture of MI-DRCT. We have divided it into two stages. Step (a) does the shallow feature extraction and alignment of shifted features. Step (b) takes the combined shallow features as input and extracts the deep features using Residual Dense Group (RDG) blocks. The deep and shallow features are then combined to produce the super-resolution image.
• Figure 5: Qualitative comparison of different super-resolution techniques on a real image captured by RootCam for a Bell Pepper plant. We can see the higher contrast in the box for MI-DRCT which enhances the root hairs. DRCT also has good contrast but slightly lower quality than the MI-DRCT.