Engineering Journals | Telecommunications Journal | Performance Testing of Composite Dielectric Patch Antenna

The IUP Journal of Telecommunications

A Novel Technique with L_o Gradient Minimization for Rain-Free Images

Article Details

Pub. Date	: Feb, 2020
Product Name	: The IUP Journal of Telecommunications
Product Type	: Article
Product Code	: IJTC40220
Author Name	: S Rakoth Kandan, N Dhanasekar
Availability	: YES
Subject/Domain	: Engineering
Download Format	: PDF Format
No. of Pages	: 13

Price

Download

Abstract

Detection and removal of rain in the image are a complicated and critical problem due to difficulty of rain and its negative effects on the image. The paper proposes a novel, efficient and simple algorithm for detection and removal of rain from the image using Lo gradient minimization that can commonly eliminate rain pixels from the image. The algorithm is easy to use and successful in spotting and removing rain of sequential images. This method does not depend on local features but as an alternative locates necessary edges globally. This relevant edge information will wait until all other information is extracted in this manner and then rain pixels are removed. Next, to improve contrast in images, intensity enhancing is done using histogram adjustment technique. To compare this result with other techniques, quality index has been measured for the rain removal image. Experimental results showed that the proposed algorithm is extremely efficient as it removes rain successfully for even light and heavy rain in the image.

Description

Different weather situations like rain, snow and fog cause problems in visual effects of the spatial or temporal domain in images (Barnum et al., 2010; and Yeh et al., 2013). However, adverse weather condition is unavoidable for any outdoor vision system. Detection of rain water drops and removal of rain from the images are essential to get perfect visualization. Rain removal process can be applied to both videos and images. It is essential to build new algorithms in a way that they are robust to weather changes. A pioneering work on detection and removal of raindrops by the meteorological approach using the global motion compensation was done by Abhishek and Sudipta (2013). A single-image-based rain streak removal framework has been proposed by formulating it as an image decomposition problem based on the Morphological Component Analysis (MCA) (Peyre et al., 2007; and Fadili et al., 2010). By exploiting PCA and SVM classifiers on the learned dictionary sets, the framework aims at automatically identifying the common rain patterns present in them, and hence rain can be removed in particular as high-frequency components (Chen and Lap-Piu, 2013). The first step is to decompose an image into low-frequency and high-frequency parts using a bilateral filter (Fu et al., 2011). The high-frequency part is next decomposed into "rain component" and "non-rain component" by dictionary learning and sparse coding. And hence, the rain component can be successfully removed from the image.