Abstract:
Vision-based intelligent systems like driver assistance and transportation should consider weather. Haze in images can endanger drivers. Real-world haze density measures image visibility and usability. In outdoor vision-based intelligent systems, haze density prediction can be useful. Haze density prediction is difficult because many scene contents look like haze. Existing methods predict image visibility or haze density using different priors and complex handcrafted features. HazDesNet is an end-to-end convolutional neural network (CNN) method for haze density prediction. HazDesNet predicts pixel-level haze density maps from hazy images. The image’s global haze density is calculated from the density map’s average after refinement and smoothing. A subjective human study is used to build a Human Perceptual Haze Density (HPHD) database of 500 real-world and 100 synthetic hazy images and their human-rated perceptual haze density scores to test HazDesNet. Our HPHD database and existing databases show that our method predicts haze density best. Besides global quantitative results, our HazDesNet can predict a continuous, stable, fine, and high-resolution haze density map.
Note: Please discuss with our team before submitting this abstract to the college. This Abstract or Synopsis varies based on student project requirements.
Did you like this final year project?
To download this project Code with thesis report and project training... Click Here
