AEA-RDCP: An Optimized Real-Time Algorithm for Sea Fog Intensity and Visibility Estimation

Applied Sciences, Journal Year: 2024, Volume and Issue: 14(17), P. 8033 - 8033

Published: Sept. 8, 2024

Sea fog reduces visibility to less than 1 km and is a major cause of maritime accidents, particularly affecting the navigation small fishing vessels as it forms when warm, moist air moves over cold water, making difficult predict. Traditional measurement tools are costly limited in their real-time monitoring capabilities, which has led development video-based algorithms using cameras. This study introduces Approximating Eliminating Airlight–Reduced DCP (AEA-RDCP) algorithm, designed address issue where sunlight reflections mistakenly recognized existing sea intensity algorithms, thereby improving performance. The dataset used experiment categorized into two types: one consisting images unaffected by another heavily influenced sunlight. AEA-RDCP algorithm enhances previously researched RDCP effectively eliminating influence atmospheric light, utilizing initial stages Dark Channel Prior (DCP) process generate image. While typically for dehazing, this employs only point generating Channel, reducing computational complexity. generated image then estimate based on threshold density estimation, maintaining accuracy while demands, allowing conditions, enhancing safety, preventing accidents.

Language: Английский

Physics-Driven Image Dehazing from the Perspective of Unmanned Aerial Vehicles

Electronics, Journal Year: 2024, Volume and Issue: 13(21), P. 4186 - 4186

Published: Oct. 25, 2024

Drone vision is widely used in change detection, disaster response, and military reconnaissance due to its wide field of view flexibility. However, under haze thin cloud conditions, image quality usually degraded atmospheric scattering. This results issues like color distortion, reduced contrast, lower clarity, which negatively impact the performance subsequent advanced visual tasks. To improve unmanned aerial vehicle (UAV) images, we propose a dehazing method based on calibration scattering model. We designed two specialized neural network structures estimate unknown parameters model: light intensity A medium transmission t. calculation errors always occur both processes for estimating parameters. The error accumulation will cause deviation fidelity brightness. Therefore, an encoder-decoder structure irradiance guidance, not only eliminates but also enhances detail restored image, achieving higher-quality results. Quantitative qualitative evaluations indicate that our outperforms existing techniques, effectively eliminating from drone images significantly enhancing clarity hazy conditions. Specifically, compared experiment R100 dataset demonstrates proposed improved peak signal-to-noise ratio (PSNR) similarity index measure (SSIM) metrics by 6.9 dB 0.08 over second-best method, respectively. On N100 dataset, PSNR SSIM 8.7 0.05

Language: Английский