Single image fog removal using COIN filters

The proposed novel algorithm helps to resolve the problem of degraded illumination and the assessment of its range in foggy weather. Like contrast restoration methods, the proposed algorithm improves the accuracy of object of interest present in the foggy image using Convergence Index Filters. The existing methods work based on the statistical measures and it may not work for the images where the scene objects are inherently similar to the atmospheric light. In the proposed algorithm the gradient energy magnitude along with its directional information is used as an additional parameter to find the objects embedded in foggy image. Some of the convergence index filters are Coin filter (CF), Adaptive Ring filter (ARF), IRIS filter (IR), Sliding Band filter (SBF). The gradient of an image can be an important source of information for texture discrimination. The property of the COIN filter works with vector data and does not depend on the contrast information. When this property is applied in fog removal, the proposed algorithm estimates the transmission map to maximize the contrast of the output image.