A study of scattered particles removal based on contrast recovery

Talib, Muhamad Lazim; Nasrudin, Mohammad Faidzul; Abdullah, Siti Norul Huda Sheikh

doi:10.1109/icaees.2016.7888073

Cited by 1 publication

(1 citation statement)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Uniform scene distorted in the image may cause noise and artifact [2]. The main activities like object recognition, tracking, and monitoring in outdoor camera system often disruption with the weather, as haze and fog [3]. Haze is tiny particles, and vapors distribute in the air that may reduce the performance of the vision system as light disturbance, and uniform scene distorted [4].…”

Section: Introductionmentioning

confidence: 99%

Uniform Distorted Scene Reduction on Distribution of Colour Cast Correction

Talib

Nasruddin

Abdullah

2019

Int. J. Adv. Sci. Eng. Inf. Technol.

View full text Add to dashboard Cite

Scene in the photo occulated by uniform particles distribution can degrade the image quality accidently. State of the art pre-processing methods are able to enhance visibility by employing local and global filters on the image scene. Regardless of air light and transmission map right estimation, those methods unfortunately produce artifacts and halo effects because of uncorrelated problem between the global and local filter's windows. Besides, previous approaches might abruptly eliminate the primary scene structure of an image like texture and colour. Therefore, this study aims not solely to improve scene image quality via a recovery method but also to overcome image content issues such as the artefacts and halo effects, and finally to reduce the light disturbance in the scene image. We introduce our proposed visibility enhancement method by using joint ambience distribution that improves the colour cast in the image. Furthermore, the method is able to balance the atmospheric light in correspondence to the depth map accordingly. Consequently, our method maintains the image texture structural information by calculating the lighting estimation and maintaining a range of colours simultaneously. The method is tested on images from the Benchmarking Single Image Dehazing research by assessing their clear edge ratio, gradient, range of saturated pixels, and structural similarity metric index. The scene image restoration assessment results show that our proposed method had outperformed resuls from the Tan, Tarel and He methods by gaining the highest score in the structural similarity index and colourfulness measurement. Furthermore, our proposed method also had achieved acceptable gradient ratio and percentage of the number of saturated pixels. The proposed approach enhances the visibility in the images without affecting them structurally.

show abstract

Section: Introductionmentioning

confidence: 99%