Scattering Phenomena: Optics of the Atmosphere . Scattering by Molecules and Particles. Earl J. McCartney. Wiley, New York, 1976. xviii, 408 pp., illus. $24.95.

A multimodal feature fusion image dehazing method with scene depth prior

“…The problem of haze degradation of images can be represented by a classical atmospheric scattering model [1], as shown in Equation (). $$$$\begin{equation}I(x) = J(x)t(x) + A(1 - t(x))\end{equation}$$$$where I ( x ) represents the hazy image acquired by the sensor, J ( x ) represents the clear image, t ( x ) represents the atmospheric transmittance, and A is the atmospheric light.…”

“…The problem of haze degradation of images can be represented by a classical atmospheric scattering model [1], as shown in Equation (1).…”

“…In this paper, three sets of ablation experiments are designed to verify the superiority of the proposed multimodal feature fusion module and FEM. (1) The original network architecture. Instead of multimodal feature fusion, a separate image feature space is used, and only convolutional layers with the same number of layers are used to replace the DFFM module and the FEM module, labelled as I; (2) the FEM module is added separately.…”

“…In the field of computer vision, this type of image degradation problem can be represented by a classical atmospheric scattering model [1]. That is, the core of the image dehazing problem is to estimate the transmittance t and the atmospheric light A .…”

New Nighttime Dehazing Algorithm Based On Dark Channel Prior Enhancement

“…Non-physical model approaches for image enhancement focus on dehazing,with techniques such as histogram equalization [9] and retinex algorithms [10][11].Tan et al [12] divided the image into small blocks and maximized the contrast of each block,while Bekaert et al [13] and Ancuti et al [14]employed a fusion-based approach that combines white balance and contrast enhancement.Choi et al [15] proposed a non-reference-aware haze density prediction model to enhance images.However,these methods lack a reasonable physical explanation for their results,and their effectiveness is not sufficient.On the other hand,physical model dehazing algorithms are based on atmospheric scattering theory [16],such as Narasimhan et al's [17] visual performance,Zhu et al's color decay preceding. [18],He et al's [19] DCP and Berman et al's haze line prior [20][21].Physical model dehazing algorithms tend to be more effective than non-physical model dehazing methods.Lastly,the advent of deep learning techniques resulted in researchers applying them to image dehazing.Shao et al [22] proposed an end-to-end domain adaptive network,Dong et al [23] developed a multiscale enhanced dehazing network,and Qu et al [24] created a multiscale enhanced dehazing network,which combines the enhanced pix2pix dehazing network into the adversarial network for image dehazing.…”