Image Acquisition and Preprocessing

Image Acquisition

Types of Sensors

• Passive Sensors:

  • Detect natural energy (e.g., sunlight) reflected or emitted by objects.
  • Examples: Cameras, multispectral, and hyperspectral sensors.
  • Uses: Land cover mapping, vegetation analysis.

• Active Sensors:

  • Emit their own signal and measure the reflected signal.
  • Examples: Radar, LiDAR (Light Detection and Ranging).
  • Uses: Topographic mapping, vegetation structure analysis.

Platforms for Data Collection

• Satellites:

  • Low Earth Orbit (LEO): 180 km to 2,000 km above Earth (e.g., Sentinel satellites).
  • Geostationary Orbit: Fixed position over one location at about 35,786 km (e.g., weather satellites).

• Aircraft:

  • Manned Aircraft: Equipped with cameras or sensors for high-resolution imagery.
  • Drones (UAVs): Used for detailed, small-area surveys.

Preprocessing

Importance of Preprocessing

• Objective: Correct and enhance raw satellite data for accurate analysis.
• Goals: Remove errors, normalize data, and prepare images for further processing.

Common Preprocessing Steps

1. Radiometric Correction:

   • Adjusts pixel values to correct sensor errors and atmospheric effects.
   • Methods:
     • Dark Object Subtraction: Corrects for atmospheric scattering.
     • Top-of-Atmosphere (TOA) Reflectance: Converts raw data to reflectance values.

2. Geometric Correction:

   • Aligns images to a map coordinate system or another image.
   • Methods:
     • Orthorectification: Corrects geometric distortions using a digital elevation model (DEM).
     • Image Registration: Aligns multiple images taken at different times or by different sensors.

3. Atmospheric Correction:

   • Removes atmospheric effects to retrieve true surface reflectance.
   • Methods:
     • Radiative Transfer Models: Simulate radiation passage through the atmosphere.

4. Cloud Masking:

   • Identifies and removes cloud-covered pixels.
   • Methods:
     • Thresholding: Uses reflectance values to differentiate clouds.
     • Automated Algorithms: Methods like Fmask using multiple spectral bands.

5. Noise Reduction:

   • Removes random noise affecting image quality.
   • Methods:
     • Filtering Techniques: Median filter, Gaussian filter.

6. Resampling:

   • Adjusts pixel size to match different datasets or improve resolution.
   • Methods:
     • Nearest Neighbor: Simplest method for categorical data.
     • Bilinear Interpolation: Averages the four nearest pixels.
     • Cubic Convolution: Uses the nearest sixteen pixels for smoother results.