Dataset DOI: 10.5061/dryad.6t1g1jxdh

Description of the data and file structure

These data were collected as part of a study on reproducible leaf shape quantification using elliptic Fourier analysis. They were generated to document the processing workflow implemented in LeafContourEFD, to compare normalization procedures, and to support downstream multivariate analyses of leaf outlines from Quercus serrata, Q. crispula, and Triadica sebifera.

Folder names indicate dataset type and taxonomic group: folder ending in quercus contain data for Q. serrata and Q. crispula, whereas folders ending in triadica_sebifera contain data for T. sebifera. These files document the major intermediate and final outputs produced by LeafContourEFD, from image selection and segmentation to contour extraction and coefficient calculation. The dataset can be reused for studies of leaf shape variation, benchmarking of image-processing and morphometric workflows, comparison of EFD normalization procedures, and reproducible reanalysis of outline-based morphometric data.

Files and variables

File: data (included in data.zip)

Description

• raw/: scanned source images of leaves.
• contour_[dataset]/: CSV files of extracted contour coordinates.
• contour_image_[dataset]/: binary contour-rendered images (contour = white, background = black).
• rois_[dataset]/: ROI overview images showing sampled leaves on original scans.
• cropped_images_[dataset]/: cropped ROI images.
• rotated_images_[dataset]/: cropped ROI images rotated by base–tip landmarks.
• binarized_images_[dataset]/: automatically generated binary leaf masks before manual correction.
• binarized_image_final_[dataset]/: final binary masks after manual correction.
• metadata_[dataset]/: per-ROI metadata in JSON and compact CSV.
• coefficients_efd_[dataset]/: raw elliptic Fourier descriptor (EFD) coefficients.
• coefficients_efd_normalized_[dataset]/: oriented true normalized EFD coefficients.
• species_name.csv: mapping table for image filenames and species names (Quercus dataset).

Dataset grouping

• folders ending in quercus: Quercus serrata and Q. crispula
• folders ending in triadica_sebifera: Triadica sebifera

Variables and abbreviations

A) contour_[dataset]/*.csv

• x: x-coordinate of contour point (pixel)
• y: y-coordinate of contour point (pixel)

B) coefficients_efd_[dataset]/*.csv and coefficients_efd_normalized_[dataset]/*.csv

• n: harmonic number (unitless integer)
• A0: x-direction constant term (pixel-scale coordinate unit)
• C0: y-direction constant term (pixel-scale coordinate unit)
• an, bn, cn, dn: Fourier coefficients for harmonic n (pixel-scale coordinate unit)

C) metadata_[dataset]/*.csv (compact metadata)

• file_absolute_path: absolute path of original source image
• file_relative_path: relative path of source image
• id: source image ID (filename stem)
• leaf_id: ROI/leaf index (usually zero-padded, e.g., 01)
• px_per_cm: scale (pixels per centimeter)
• base_x, base_y: rotated-base landmark coordinate (pixel)
• tip_x, tip_y: rotated-tip landmark coordinate (pixel)

D) metadata_[dataset]/*.json (detailed metadata)

• metadata_version: metadata schema version
• source:
  • absolute_path: original local absolute path
  • relative_path: relative path within project/dataset context
  • image_id: source image identifier
  • roi_index: ROI index
• scale:
  • px_per_cm: pixels per centimeter
  • unit: calibration unit (cm)
  • scale_factors: applied layer scaling factors [y_scale, x_scale]
  • dpi: dots per inch (if available)
• roi:
  • polygon_yx: ROI polygon vertices in (y, x) pixel coordinates
  • bbox_ymin_ymax_xmin_xmax: ROI bounding box (ymin, ymax, xmin, xmax) in pixels
  • corners_yx: bounding-box corners in (y, x) pixel coordinates
  • slice_indices: Optional internal viewer state (dims.current_step) recorded at ROI creation; not used as a primary analytical variable.
• rotation:
  • angle_deg: rotation angle (degrees)
  • original_size: pre-rotation image size (pixels)
  • rotated_size: post-rotation image size (pixels)
• landmarks:
  • points_layer_name, points_n, points_labels
  • base_original, tip_original: original landmark coordinates
  • base_rotated, tip_rotated: rotated landmark coordinates
• binarization:
  • method: binarization method used (Otsu or SAM2)
  • threshold: threshold value used for Otsu binarization; null when method = "SAM2".
  • manually_edited: whether mask was manually corrected (true/false)
• contour:
  • points: number of contour points
  • area: contour area (pixel²)
• meta:
  • created_time: metadata creation timestamp (ISO 8601)
  • cropped_from, face_color_type, border_color_type
• processing_history: ordered processing log (binarization, contour extraction, etc.)

Missing data notation

• Missing values in CSV files are represented as blank cells.
• Missing values in JSON files are represented as null.

Code/software

No specialized proprietary software is required to view the data files.

The dataset consists of standard formats (.csv, .json, .png, .jpg) that can be opened with common free software (e.g. LibreOffice or any text editor for CSV/JSON, and standard image viewers for PNG/JPEG).

For reproducible re-analysis of contour extraction and EFD outputs, we used LeafContourEFD (open-source, Python-based GUI).

If needed, user can reproduce processing with Python (e.g. via pandas for tabular data and image-processing libraries such as OpenCV), but this is not required for basic data access.

Access information

Other publicly accessible locations of the data:

• Not applicable

Data was derived from the following sources:

• Not applicable. The data were generated by the authors from original scanned leaf images and subsequent processing.