Atmospheric light estimation through color vector orthogonality for image dehazing
Lingzhao Kong, Dagang Jiang, Xin Liu, Yu Zhang, Kaiyu Qin · Optical Engineering · 2024
Atmospheric light estimation is fundamental for image dehazing. Unlike previous research on the atmospheric light estimation, we reveal a vector orthogonality in the classical atmospheric color degradation model for the first time. It is demonstrated that atmospheric light vector is orthogonal to the normal vectors of color degradation plane. Furthermore, an estimation method for direction and magnitude of atmospheric light vector is proposed based on the vector orthogonality. The proposed method can accurately estimate the atmospheric light and improve the quality of image dehazing with a short constant processing time even if the image resolution is increasing. We provide an efficient approach for estimating atmospheric light for high-definition image dehazing applications, such as traffic surveillance, environmental monitoring, and aerial photography.