This article was automatically translated from the original Turkish version.

Dark Channel Prior Algorithm

+1 More

Fog is a common issue in outdoor photographs. Atmospheric dust, moisture, and other particles scatter and absorb light, reducing contrast and color accuracy in the image. This situation leads to the loss of detail in distant objects road. Traditional methods require multiple images or additional hardware, while single-image dehazing techniques offer an innovative approach to solving this problem.

The Dark Channel Prior (Darkness Canal Prior) algorithm was initially developed by Kaiming He and his team to recover clear images from hazy ones.

Dark Channel Prior (DCP)

DCP is based on a statistical property observed in non-foggy outdoor images. He and colleagues (2009) found through analysis of over 5,000 non-foggy images that, except in bright regions, most local patches contain at least one color channel (red, green, or blue) with very low pixel intensity. This low intensity typically originates from shadows, dark colorful objects, or non-bright surfaces. Mathematically, the dark channel is defined as:

Jdark(x)=minc∈r,g,b(miny∈Ω(x)Jc(y))

Steps of the Original Dark Channel Prior Algorithm

The original algorithm developed by He and colleagues (2009) performs dehazing in five key steps.

1- Estimation of Atmospheric Light

Atmospheric light (A) is typically estimated from the brightest region of the image. He and his team select the brightest 0.1% of pixels in the dark channel and compute A as their average. This method yields a more robust result than directly selecting the brightest pixel, as it reduces errors caused by noise or bright objects.

2- Calculation of the Dark Channel

The image is divided into local patches (e.g., 15x15 pixels). For each patch, the minimum intensity across each color channel is found, and the lowest of these minima is selected as the dark channel:

Idark(x)=minc∈r,g,b(miny∈Ω(x)Ic(y))

This operation, Marcel van He’s fast minimum filter is implemented efficiently, and computational complexity scales linearly with image size.

3- Initial Estimation of the Transmission Map

Using the dark channel, the transmission map is calculated as follows:

t(x)=1−ω∗(minc(miny∈Ω(x)Ic(y)/Ac))

4- Refined Transmission Map

The initially estimated transmission map is often noisy and has blurred edges. Therefore, it is refined using techniques such as soft matting or guided filtering. Soft matting minimizes the following cost function:

E(t)=tTLt+λ(t−that)T(t−that)

5- Recovery of the Clear Image

In the final step, the clear image is recovered from the hazy image using the following formula:

Advantages and Limitations of the Algorithm

The dark channel prior algorithm delivers impressive results in fog removal from a single image and requires no additional hardware. However, it has certain limitations:

Bright Areas: In homogeneous and bright regions such as the sky, the dark channel assumption may not hold. In these areas, pixel intensity is not close to zero, which can lead to color distortion or noise.
Similar Color Problem: If objects in the scene have colors similar to the atmospheric light (for example, a white building), the algorithm may make incorrect estimates.
Computational Complexity: Refinement steps such as soft matting can be computationally intensive.

Application Areas

The dark channel prior algorithm can be applied in various fields:

Photography: Enhancing foggy outdoor photographs.
Autonomous Vehicles: Improving visibility for camera-based systems in foggy conditions.
Satellite Imaging: Clarifying cloudy or foggy satellite images.
Security Systems: Increasing the effectiveness of security cameras in foggy environments.

DCP Algorithm Implementation Code

Example Output:

Application of the Dark Channel Prior algorithm (Prepared and edited by: Beyza Nur Türkü)

Author Information

AuthorBeyza Nur TürküDecember 20, 2025 at 7:18 AM

Discussions

No Discussion Added Yet

Start discussion for "Dark Channel Prior Algorithm" article

View Discussions

Dark Channel Prior (DCP)
Steps of the Original Dark Channel Prior Algorithm
- 1- Estimation of Atmospheric Light
- 2- Calculation of the Dark Channel
- 3- Initial Estimation of the Transmission Map
- 4- Refined Transmission Map
- 5- Recovery of the Clear Image
Advantages and Limitations of the Algorithm
Application Areas
DCP Algorithm Implementation Code