跳到主要内容

MCTP: Unified Multi-Modal Phenotyping Data Processing Platform

· 阅读需 3 分钟
Liangchao Deng
Liangchao Deng
Ph.D. Candidate @ SHZU @CAS-Cemps

Platform Overview​

MCTP (Multi‑modal Crop Trait Processing) is a unified data processing platform designed for plant phenotyping workflows. It brings hyperspectral, LiDAR, RGB, and thermal imaging into a single pipeline with consistent GUI experiences, reproducible parameter tuning, batch processing, and standardized outputs—making multi‑modal analysis easier to manage and scale.

MCTP is a self-developed field walking phenotyping platform by Shufeng Bio, and I was responsible for system optimization and data processing and analysis during the development process. mctp interface

Core Modules and Capabilities​

1) Hyperspectral: HyperVis Module​

Hyperspectral module placeholder

  • ENVI data parsing: HDR/SPE ingestion with wavelength parsing.
  • Auto band detection: Multi band indices inferred from wavelength metadata.
  • Index calculation and masks: NDVI with NIR thresholding and glint percentile masks.
  • Spectral statistics: Leaf‑region mean spectrum (CSV) and summary metrics (JSON).
  • Tabbed visualization: RGB quicklook, NDVI, mask comparison, spectral curves, and stats.
  • Batch processing: Directory‑level processing for multiple HDR/SPE pairs.

2) LiDAR: Point Cloud Fusion and Trait Quantification​

LiDAR module placeholder LiDAR module placeholder

  • Multi‑format I/O: PLY / LAS / LAZ / TXT support.
  • Geometry processing: RANSAC ground rebasing, voxel downsampling, cropping, height coloring.
  • Canopy and plant traits: Coverage, voxel occupancy, convex hull volume, H10/H50/H90, etc.
  • Clustering and tuning: DBSCAN segmentation with an interactive SegTuner window.
  • Exports: Cropped point clouds and JSON trait reports.

3) RGB: Robust Plant Segmentation and Group Metrics​

  • Multi‑index vegetation fusion: ExG + CIVE + VDI with joint thresholding.
  • Glare suppression and noise cleanup: Handles water reflections, labels, and borders.
  • Thin‑plant robustness: Adaptive morphology and skeleton enhancement.
  • Plant instance labeling: Watershed + connected components for per‑plant IDs.
  • Exports: Overlay images, JSON metrics, and per‑plant CSV tables.

4) Thermal: Temperature/HSV Dual‑Channel Segmentation​

Thermal module placeholder

  • BMP + DDT workflow: Temperature matrix loading, flipping, calibration.
  • Suggested thresholds: Otsu + percentiles for quick tuning.
  • Interactive sliders: Thresholds, HSV range, morphology parameters.
  • Multi‑view preview: Overlay / heatmap / plants‑only views.
  • Exports: PNG, NPY, CSV, and JSON reports.

Unified Workflow Highlights​

  • Consistent UI experience across all four modalities with control panels, logging, and previews.
  • Batch processing support for hyperspectral and RGB pipelines; fast parameter reuse for LiDAR and thermal.
  • Parameter‑tuning friendly with sliders, suggested thresholds, and real‑time previews.
  • Structured outputs in JSON/CSV for downstream statistics and data sharing.

Typical Use Cases​

  1. Greenhouse or field campaigns capturing hyperspectral, LiDAR, RGB, and thermal data per sample.
  2. Unified parameter management for consistent thresholding and QC across modalities.
  3. Batch analysis and archiving with standardized outputs for trait‑yield modeling.

Roadmap​

  • Cross‑modal registration and temporal analytics
  • Unified project configuration templates
  • Cloud‑based batch processing and reporting

If you want to try MCTP or access sample datasets, feel free to reach out.

CN Community

EN Community

More