HyperVein: A Dataset for Human Vein Detection from Hyperspectral Images

HyperVein Dataset:

This folder contains a subset of hyperspectral images from the "HyperVein" dataset, which originally consists of 400 images of human hands and arms collected from 100 subjects.

Folder Contents

1. Hyperspectral Image Dataset

The dataset includes 8 hyperspectral images captured from 2 volunteer participants.
For each participant, images of both the left and right hand and arm are provided.

Naming Convention

Files are named using the following format:

SXXX_L_A → Left Arm image
SXXX_L_H → Left Hand image
SXXX_R_A → Right Arm image
SXXX_R_H → Right Hand image

Where:

XXX represents the subject ID ranging from 001 to 100.

Example:

S019_L_A.bil → Subject 019, Left Arm
S019_L_H.bil → Subject 019, Left Hand
S019_R_A.bil → Subject 019, Right Arm
S019_R_H.bil → Subject 019, Right Hand

For each hyperspectral image:

The image data is provided in Band Interleaved by Line (.bil) format.
The corresponding header file (.hdr) is also included.

2. Cropped RGB Images and Ground Truth Annotations

This folder also contains:

Cropped RGB images corresponding to each hyperspectral image
Annotated ground truth vein masks
Image Dimensions
Hand images: 512 × 512
Arm images: 336 × 496

These images represent the selected Region of Interest (ROI) used for the vein detection experiments.

The ground truth images are scaled between 0 and 255 for visualization purposes.

3. MATLAB Example Script

An example MATLAB script named ReadHSI.m is included for reading and displaying the hyperspectral images.

Before running the script, install the MATLAB Hyperspectral Image Processing Toolbox.

Notes:

This folder contains only a subset of the original dataset.
The dataset is intended for research and educational purposes related to hyperspectral imaging and vein detection.