跳到主要内容

Stereo Camera Leaf Measurement System Tutorial

Overview​

This is a high-precision stereo vision system specifically designed for leaf morphological measurements. The system uses your calibration parameters for image correction, providing millimeter-level depth measurement accuracy, suitable for leaf research within a 30cm height shooting box.

Key Features​

  • High-precision Image Correction: Distortion correction and stereo rectification using calibrated intrinsic matrices
  • Accurate Depth Measurement: Millimeter-level depth accuracy within 5-30cm range
  • Leaf-optimized Algorithms: Stereo matching specifically optimized for plant leaf characteristics
  • Corrected Image Capture: Save high-quality stereo image pairs after correction
  • Precise Depth Data: Suitable for leaf thickness and morphological analysis

Quick Start​

1. Access the Application​

Visit in your browser: /app/stereo

2. Hardware Requirements​

  • Stereo Camera System: Your built stereo camera supporting side-by-side format output
  • Shooting Box: 30cm height shooting environment
  • Browser: Modern browsers like Chrome, Firefox, Safari (WebAssembly support required)
  • Network Environment: HTTPS or localhost (camera permission requirements)
  • OpenCV.js: Automatically loaded by the system for high-precision image processing

Detailed Usage Steps​

Step 1: Device Setup​

  1. Connect Stereo Camera

    • Ensure the stereo camera is properly connected to the computer
    • Camera should output side-by-side format stereo images

  2. Select Camera Device

    • Choose your stereo camera from the "Camera Device" dropdown menu
    • System automatically remembers the last selected device

  3. Set Resolution

    • Width: 1280 pixels (total stereo width, 640px per eye)
    • Height: 480 pixels (per-eye resolution)
    • These parameters match your calibration data, do not modify arbitrarily

Step 2: Start Camera​

  1. Click "Start Camera" Button

    • First use will request camera permission, click "Allow"
    • Success will display real-time video stream

  2. Check Image Display

    • Original Stream: Displays raw stereo video stream
    • Left Eye: Distortion-corrected and stereo-rectified left eye image
    • Right Eye: Distortion-corrected and stereo-rectified right eye image
    • Depth Map: Shows "Image Corrected" prompt, waiting for depth calculation

Step 3: Precise Depth Calculation​

  1. Click "Calculate Depth" Button

    • System uses OpenCV's StereoBM algorithm for high-precision stereo matching
    • Parameter settings optimized for leaf features
    • Generates precise depth map within 5-30cm range

  2. Depth Map Grayscale Rendering Principle

    • System automatically detects nearest distance and farthest distance in the scene
    • Normalized rendering based on actual depth value range
    • Black (0): Nearest distance in scene or invalid depth regions
    • White (255): Farthest distance in scene
    • Intermediate Grayscale: Linearly distributed by depth value

    Depth Value Calculation Formula:

    Grayscale Value = (Current Depth - Nearest Depth) / (Farthest Depth - Nearest Depth) × 255
    Actual Depth = Nearest Depth + (Grayscale Value / 255) × (Farthest Depth - Nearest Depth)

    Advantages:

    • Full utilization of 0-255 grayscale range
    • Adaptive to scene depth distribution
    • Maximized depth detail contrast

Step 4: Image Capture​

  1. Set Sample ID

    • Enter descriptive name in the "Sample ID" input box
    • Examples: "Desktop Objects", "Indoor Scene", etc.

  2. Capture Corrected Stereo Image Pair

    • Click "Capture Image Pair" button
    • System saves high-quality images after distortion correction and stereo rectification
    • File naming format: SampleID_stereo_001_left_rectified.png, SampleID_stereo_001_right_rectified.png
    • These images can be directly used for subsequent scientific analysis

  3. Capture Precise Depth Map (Recommended)

    • First execute depth calculation
    • Click "Capture Depth Map" button
    • Save high-precision depth visualization image: SampleID_depth_001_precision.png
    • Depth map contains precise depth information within 5-30cm range

Step 5: Data Download​

  1. View Capture List

    • Check all captured images in the "Capture Data" panel
    • Can download each file individually

  2. Batch Download

    • Click "Download ZIP" button
    • System packages all captured images into ZIP file
    • File name: leaf_stereo_captures.zip
    • Contains corrected stereo image pairs and precise depth maps

Technical Specifications and Parameters​

Calibration Parameters (Built-in)​

System uses your precise calibration parameters:

Left Camera Intrinsic Matrix:

fx: 526.36, fy: 527.67, cx: 312.51, cy: 257.35
Distortion Coefficients: [-0.0356, 0.1847, 0, 0, 0]

Right Camera Intrinsic Matrix:

fx: 528.81, fy: 529.73, cx: 319.85, cy: 259.80
Distortion Coefficients: [-0.0274, 0.1308, 0, 0, 0]

Stereo Calibration Parameters:

  • Baseline Distance: 59.936mm (high precision)
  • Rotation Matrix R: Corrected inter-camera rotation relationship
  • Translation Vector T: [-59.936, 0.006, 0.957]mm

Leaf Measurement Optimization Parameters​

StereoBM Algorithm Settings:

  • Number of Disparities: 64 pixels (suitable for 30cm shooting distance)
  • Block Size: 15×15 pixels (balances accuracy and detail)
  • Uniqueness Ratio: 10% (ensures matching reliability)
  • Speckle Filtering: Window 50 pixels, range 2 pixels

Depth Measurement Range:

  • Minimum Depth: 50mm (5cm)
  • Maximum Depth: 300mm (30cm)
  • Depth Accuracy: Sub-millimeter level (theoretical accuracy ~0.5mm)

Troubleshooting​

Common Issues​

1. Camera Cannot Start

  • Check if camera is properly connected
  • Confirm browser has granted camera permission
  • Try refreshing page to re-authorize

2. Abnormal Image Display

  • Confirm camera outputs side-by-side format
  • Check if resolution settings are correct
  • Try different camera devices

3. Abnormal Depth Map Display

  • Ensure leaf surface has sufficient texture features
  • Check if lighting in shooting box is uniform and soft
  • Avoid strong reflections or shadows
  • Ensure leaves are within 5-30cm measurement range

4. Insufficient Depth Measurement Accuracy

  • Check if camera is stably fixed
  • Ensure leaf surface is clean, free of water droplets or dust
  • Use diffuse light sources, avoid point light sources
  • Ensure leaves are flat, minimize bending deformation

5. Poor Image Correction Effect

  • System uses your calibration parameters, check calibration quality if issues occur
  • Ensure camera position is consistent with calibration
  • Check if lenses are clean

Leaf Measurement Optimization​

Improve Measurement Accuracy:

  • Use uniform ring LED light sources
  • Ensure leaf surfaces are dry and clean
  • Maintain stable shooting environment temperature
  • Use tripod to fix camera system

Leaf Placement Techniques:

  • Lay leaves flat on neutral background
  • Avoid leaf overlap or bending
  • Ensure main parts of leaves are within 15-25cm depth range
  • Use background materials with moderate contrast

Data Quality Control:

  • Shoot multiple angles for each sample
  • Record environmental conditions during shooting
  • Regularly check system calibration accuracy
  • Save original corrected images for subsequent analysis

Technical Specifications​

Supported Image Formats​

  • Input: Real-time video stream (stereo side-by-side, 1280×480)
  • Output: PNG format high-quality images
  • Depth Data: Visualized PNG images (future support for raw depth data export)

Browser Compatibility​

  • Chrome 90+ (recommended, best WebAssembly performance)
  • Firefox 85+
  • Safari 14+
  • Edge 90+

System Requirements​

  • Memory: Recommended 8GB+ (OpenCV processing requires)
  • Processor: WebAssembly support, recommended multi-core CPU
  • Network: HTTPS or localhost access
  • Graphics Card: Hardware acceleration support (optional, improves performance)

Usage Tips​

Leaf Measurement Best Practices​

  1. Leaf Preparation

    • Select healthy, complete leaf samples
    • Clean leaf surfaces, remove dust and water droplets
    • Ensure leaves are relatively flat, minimize bending

  2. Shooting Environment Setup

    • Use ring LED light sources for uniform illumination
    • Choose neutral background colors (light gray or white)
    • Control environmental temperature and humidity to avoid leaf deformation
    • Eliminate vibration and airflow effects

  3. Measurement Operation Standards

    • Wait for complete system initialization after startup
    • Confirm good image correction effect
    • Take multiple measurements per sample and average
    • Record detailed measurement conditions and parameters

  4. Data Management

    • Use descriptive sample IDs for naming
    • Establish standardized file naming conventions
    • Regularly backup measurement data
    • Record metadata related to each sample

Data Management​

  1. File Naming

    • Use descriptive sample IDs
    • Classify by date or scene
    • Maintain naming consistency

  2. Data Backup

    • Regularly download ZIP files
    • Establish local backup strategy
    • Record shooting parameters and conditions

Technical Support​

If you encounter problems or need technical support, please:

  1. Check browser console error messages
  2. Confirm hardware connections and settings
  3. Try basic troubleshooting steps
  4. Record specific error phenomena and environmental information

Update Log​

Current Version Features (v2.0 - Leaf Measurement Special Edition)​

  • High-precision image correction (using your calibration parameters)
  • OpenCV StereoBM depth calculation
  • 5-30cm precise depth measurement
  • Leaf-optimized stereo matching algorithm
  • Corrected image capture
  • Precise depth map saving
  • Batch data management

Planned Features​

  • Raw depth data export (CSV/JSON format)
  • Automatic leaf contour extraction
  • Leaf thickness distribution analysis
  • Multi-spectral depth fusion
  • Real-time depth measurement display

Leaf Research Applications​

  • Morphological Analysis: Leaf thickness distribution, surface texture
  • Growth Monitoring: 3D changes during leaf development process
  • Variety Comparison: Morphological differences between different varieties
  • Environmental Response: Leaf morphological responses to environmental conditions

Author: Liangchao Deng, Ph.D. Candidate, Shihezi University / CAS-CEMPS
This tutorial applies to Leaf Measurement Stereo Camera System v2.0 Optimized for 30cm shooting box environment and leaf research