Real-Time Image Dehazing Using Optimized Dark Channel Prior with Guided Filtering and Adaptive Enhancement

Nguyen Thu Trang

Abstract: Image dehazing is a persistent challenge in computer vision, especially for real-time applications such as autonomous vehicles and surveillance systems. Traditional methods often struggle with computational speed and detail retention. This study introduces an optimized Dark Channel Prior (DCP) algorithm enhanced with guided filtering and adaptive brightness-contrast adjustments. The algorithm uses vectorized morphological operations for dark channel computation, an intensity-based approach for atmospheric light estimation, and guided filtering for transmission refinement. Performance evaluations across image resolutions from 0.4MP to 17.1MP demonstrated significant improvements in processing time and image quality. The proposed method enables real-time dehazing for lower-resolution images and near-real-time processing for higher resolutions, making it practical for real-world computer vision tasks.

Keywords: computer vision, dark channel prior, guided filtering, real-time image processing, image dehazing

How to Cite?: Nguyen Thu Trang, "Real-Time Image Dehazing Using Optimized Dark Channel Prior with Guided Filtering and Adaptive Enhancement", Volume 14 Issue 7, July 2025, International Journal of Science and Research (IJSR), Pages: 1297-1301, https://www.ijsr.net/getabstract.php?paperid=SR25717062457, DOI: https://dx.doi.org/10.21275/SR25717062457