TBNet: Three branch network for image dehazing via subspace projection
Jie Huang, Wenjun Cai, Xie Yang · 2023
The advancement of modern society is always followed by environmental pollution, and the haze phenomenon is becoming more common, having a significant impact on our lives. Haze is a typical atmospheric phenomena caused by small floating particles in the air such as dust and smoke. These floating particles absorb and scatter light significantly, lowering the quality of acquired photos. Many practical applications, such as video surveillance, remote sensing, and autonomous driving, are jeopardized by haze, and advanced computer vision tasks, such as detection and recognition, are difficult to complete. As a result, image defogging has become an increasingly relevant technique with great research value while also being a tough topic. In this paper, we take this as a starting point and design the TBNet: Three branch network with subspace projection as an introduction point to implement image defogging, and achieve satisfactory results on the Fog-Cityscpace dataset.