A new deep learning architecture for dehazing of aerial remote sensing images

Kalra, Abhi; Sequeira, Aaron; Manjunath, Aditya; Lal, Shyam; Raghavendra, B. S.

doi:10.1007/s11042-022-13122-5

Cited by 4 publications

(10 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Kalra et al proposed a novel deep learning architecture for dehazing of aerial remote sensing images [26]. The authors introduced an end-to-end deep learning network (EEDNet) that efficiently dehazed aerial remote sensing images by directly computing the relationship between hazed and dehazed images.…”

Section: Related Workmentioning

confidence: 99%

“…Extensive comparisons are performed among various state-of-the-art image dehazing methods, including PDL [14], EEDNet [26], DeHRNet [27], TSDNet [12], Dehaze-Former [9], DDNet [11], CPAD-Net [29], PFBN [13], and the proposed DSSCNet with different loss functions (MSE Loss, CL, and RL). The evaluations are performed on RESIDE) dataset [31] using various visual and quantitative assessments.…”

Section: Performance Analysismentioning

confidence: 99%

“…The proposed DSSCNet with RL yields the most visually compelling dehazing result, showcasing its effectiveness in addressing haze removal challenges and preserving important image details. [14], (c) EEDNet [26], (d) DeHRNet [27], (e) TSDNet [12], (f) DehazeFormer [9], (g) DDNet [11], (h) CPAD-Net [29], (i) PFBN [13], (j) Proposed DSSCNet with MSE Loss, (k) Proposed DSSCNet with CL, and (l) Proposed DSSCNet with RL standing. In Fig.…”

Section: B Hyperparameters Of Proposed Dehazing Networkmentioning

confidence: 99%

“…Similarly, Figs. 3(c) and 3(d) depict the outputs of EEDNet [26] and DeHRNet [27], respectively, both displaying moderate darkening, edge, and texture degradation, with differing levels of spectral and spatial information preservation. Fig.…”

Section: B Hyperparameters Of Proposed Dehazing Networkmentioning

confidence: 99%

“…Figs. 4(b) to 4(f) demonstrate the outputs of PDL [14], EEDNet [26], DeHRNet [27], TSDNet [12], and DehazeFormer [9], respectively. These methods exhibit over restoration with remaining haze, along with varying levels of edge and texture degradation.…”

Section: B Hyperparameters Of Proposed Dehazing Networkmentioning

confidence: 99%

See 4 more Smart Citations

DSSCNet: Deep Custom Spatial and Spectral Consistency Layer-Based Dehazing Network

Kaur,

Singh,

Kumar

et al. 2024

IEEE Access

View full text Add to dashboard Cite

Poor weather conditions, such as haze, fog, and smog, present significant challenges in capturing clear and visually appealing images. To address this issue, we propose a Deep Custom Spatial and Spectral Consistency Layer-based Dehazing Network (DSSCNet) that effectively removes haze from images while preserving important spatial and spectral details. The network architecture includes a custom Haze Removal Layer (HRL), convolutional layers with ReLU activation, pooling layers, skip connections, and a custom Spatial and Spectral Consistency Layer (cSSCL). HRL estimates atmospheric light and transmission maps to generate an intermediate haze-free image. The proposed loss function combines Mean Squared Error (MSE) loss with a Consistency Loss (CL) to encourage content preservation during dehazing. The network is trained using the Adam optimizer to optimize the loss function, resulting in a powerful dehazing network capable of producing visually realistic and haze-free images. Extensive experimental results demonstrate the effectiveness and superiority of DSSCNet when compared to existing dehazing techniques. The outcomes highlight that DSSCNet outperforms competitive models in terms of various performance metrics, including Contrast gain (c g ), new visible edges (e), new edge gradients (r), Peak Signal-to-Noise Ratio (PSNR), and Structural Similarity Index (SSIM). These improvements amount to an average enhancement of around 1.27%, 1.12%, 1.18%, 1.21%, and 1.24%, respectively.

show abstract

Section: Related Workmentioning

confidence: 99%