Deep learning for dehazing: Comparison and analysis

TL;DR

This paper compares a deep learning-based dehazing method, Dehazenet, with traditional approaches, analyzing its performance and limitations in estimating haze removal from images.

Contribution

It provides a comparative analysis of Dehazenet against traditional methods, highlighting its strengths in transmission estimation and its limitations due to the underlying imaging model.

Findings

Dehazenet accurately estimates transmission maps.

The method shares limitations with other approaches due to the Koschmieder model.

Deep learning improves certain aspects of dehazing but is constrained by the imaging model.

Abstract

We compare a recent dehazing method based on deep learning, Dehazenet, with traditional state-of-the-art approaches , on benchmark data with reference. Dehazenet estimates the depth map from transmission factor on a single color image, which is used to inverse the Koschmieder model of imaging in the presence of haze. In this sense, the solution is still attached to the Koschmieder model. We demonstrate that the transmission is very well estimated by the network, but also that this method exhibits the same limitation than others due to the use of the same imaging model.