"Combining multispectral and high-resolution 3D imaging for leaf vein segmentation and density measurement,“ (2025)

.-H. Liao and S. Zhang, “Combining multispectral and high-resolution 3D imaging for leaf vein segmentation and density measurement,“ Frontiers in Plant Science, (2025)

Abstract

Accurate leaf vein segmentation and vein density (VLA) measurement are crucial for understanding plant physiology. Traditional 2D imaging techniques often require labor-intensive and destructive processes, such as leaf flattening or chemical clearing, thereby limiting their practicality for high-throughput applications. In this study, we present a novel framework that integrates multispectral and high-resolution 3D imaging to enhance leaf vein segmentation and VLA measurement. By leveraging digital fringe projection, our system captures grayscale, multispectral, and 3D topographical data within a unified coordinate system. The integration of 3D information improves vein detection, particularly in low-contrast regions, while also enabling direct and accurate measurements of leaf area, vein length, and VLA. However , this approach also introduces some false positives in vein segmentation due to mesophyll surface variability. Despite these challenges, our high-resolution 3D imaging method shows significant potential for non-invasive phenotyping and trait assessment in complex, unstructured environments.

"Precise calibration for error detection and correction in material extrusion additive manufacturing using digital fringe projection" (2025)

W. Keller, J. Girard, M. W. Goldberg, and S. Zhang, “Precise calibration for error detection and correction in material extrusion additive manufacturing using digital fringe projection,“ Measurement Science and Technology, 36(2), 025203 (2025)

Abstract

This work proposes a novel calibration method to align the coordinate system of a digital fringe projection system and a material extrusion additive manufacturing machine for in-process error detection and correction. A calibration geometry is printed and measured with the DFP system. Three-dimensional and two-dimensional data processing techniques are then implemented to estimate a rigid transformation matrix between the coordinate systems. The proposed calibration method and geometry are designed to run every time before the start of the manufacturing process. This removes inaccuracies caused by processes in additive manufacturing such as homing and auto-leveling, and allows the transformation to have a high accuracy. The proposed calibration method has achieved a sub-millimeter accuracy with an average root mean square error (RMSE) of 0.028 mm. The proposed method was further used for process monitoring to detect and correct sub-layer under-extrusion errors on a layer-by-layer basis. This error correction process improved the average RMSE by a factor of 1.7 for a simple shape error.