An End-to-End System for Unmanned Aerial Vehicle High-Resolution Remote Sensing Image Haze Removal Algorithm Using Convolution Neural Network

Abstract

An end-to-end image dehazing method based on convolution neural network is presented to solve the problem in which Unmanned Aerial Vehicle (UAV) high-resolution remote sensing images have reduced image sharpness due to haze. First, the original atmospheric scattering model is adapted to get an end-to-end dehazing model. Then, several unknown parameters are unified into one parameter, and the unknown parameter is estimated by using a multiscale convolution neural network. Finally, the parameter estimates are incorporated into the dehazing model to get a haze-free image. For the no reference image dataset, we first train the network using existing datasets, and then the network is trained using a self-built dataset. In this paper, the haze removal effect for different types of unmanned remote sensing images is tested and compared with those of the main dehazing algorithms. The experiments show that the algorithm in this paper has different degrees of improvement regarding its visual effect and objective indicators. INDEX TERMS End to end image dehazing; Deep learning; Multiscale convolutional neural network; Remote sensing image Yufeng Li, PHD, professor and master's supervisor at the School of electronic information engineering, Shenyang University of Aeronautics and Astronautics. He graduated from Northern Traffic University in July 1991, majoring in communication engineering. In March 2002, he graduated from Jilin University, majoring in communication and information system. In

A saliency guided remote sensing image dehazing network model

“…To synthesize hazed remote sensing images, the atmospheric scattering model (ASM) [30] described in formula () is employed in this work. Inspired by Li et al [1]. We set the atmospheric light value A between [0.6, 1.0] and the transmittance $$t(x)$$between [0.35, 0.65] to generate a random number and substitute into formula ().…”

“…Inspired by Li et al [1]. We set the atmospheric light value A between [0.6, 1.0] and the transmittance t (x)between [0.35, 0.65] to generate a random number and substitute into formula (1). Finally we resize these images to 400 × 600 and convert them to Portable Network Graphics format.…”

“…However, under a bad weather condition such as haze, rain, snow, or sand dust, images captured with remote sensing satellites or unmanned aerial vehicle (UAV) are usually suffered from serious degradation, such as colour distortion, blurring, low contrast and which results in a poor visibility. Hence, it is highly valuable to improve the visual quality of remote sensing images captured in bad weather conditions [1].…”

A multimodal feature fusion image dehazing method with scene depth prior

“…The dataset contains 13,990 hazed images, which were generated from clear images of the indoor depth dataset NYU2 [34] and Middlebury [35]. Each clear image generates 10 synthetic hazed images according to the atmospheric scattering model, with global atmospheric light values taking values between 0.7 and 1.0 [37] and atmospheric scattering coefficients chosen uniformly at random between 0.6 and 1.8 [38]. The RESIDE dataset consists of two main components, the indoor training set (ITS) and the outdoor training set (OTS).…”

Uav Remote Sensing Image Dehazing Based On Saliency Guided Two-Scaletransmission Correction

“…However, most dehazing results on UAV images have color distortions. Recently, several dehazing models based on deep learning are proposed [6][7] [8]. But when applied to UAV images, these models often hold problems of texture loss in highlight regions due to insufficient haze removal (Fig.…”