Abstract

Haze removal from satellite images is vital for various outdoor applications. However, existing techniques often lack the necessary knowledge to effectively restore hazy satellite images. They tend to rely on attributes with constant values, resulting in suboptimal dehazing outcomes. This review paper presents a structured overview of well-known haze removal techniques, shedding light on their limitations. To address these drawbacks, this research leverages advanced Deep Learning Convolutional Neural Networks (DLCNN) to enhance satellite image dehazing. The network is trained and tested with Laplacian and Gaussian pyramid-based features, which are used to modify Multiple Exposure Fusion (MEF) properties, enabling precise enhancement. The proposed DLCNN-MEF technique outperforms state-of-the-art approaches in terms of various parameters, including PSNR, SSIM, and MSE values for result comparison.

Description