Adrishyam: Dehazing Images with CV

How Adrishyam uses Dark Channel Prior?

Feb 4, 2026

Ever snapped a tiger in Ranthambore or Delhi’s skyline, only to find haze turning your photo into a ghostly blur? Adrishyam is a Python package that implements the Dark Channel Prior algorithm to dehaze images—restoring lost details with physics, not deep learning.

Why Adrishyam?

  • Wildlife Photography: Recovers textures in foggy dawn shots—no more ghost cats.
  • Autonomous Vehicles: Clears sensor data for better lane detection in smog.
  • Surveillance: Real-time frame processing for actionable intelligence in humid or polluted environments.

Install & Use

pip install adrishyam
from adrishyam import dehaze_image
 
dehaze_image(
    input_path="path/to/hazy.jpg",
    output_dir="path/to/clarity"
)

How It Works

  • Dark Channel Prior: Finds low-intensity patches to estimate haze.
  • Transmission Map: Calculates how much light survives.
  • Guided Filtering: Refines the haze mask for sharp results.

Results

  • original.png: Your input
  • dehazed.png: Restored clarity
  • result.png: All steps visualized

Real-World Impact

  • Conservation: Better species ID in camera trap images
  • Traffic: Enhanced visibility for safer roads
  • Heritage: Clearer documentation of monuments

Performance

Processes a 1920×1080 image in ~2-3 seconds on mid-range hardware. No GPU required.

Try It

Adrishyam proves that classical computer vision can still solve real problems—no training data, just solid algorithms.