Classification of hazy and non-hazy images

摘要

For defogging of pictures, initially it is investigated that if the picture is foggy or clear and for this someone cannot depend upon view of human, so for deciding the picture state as foggy or clear must be comprehended first. In this proposed work, 19 classification techniques are applied to judge whether a picture is foggy or clear. Based on characteristic difference of foggy and clear pictures, the five parameters which are area, mean, min intensity, max intensity, standard deviation are used for training the classification system. Some of the classification techniques provide very good accuracy and can be used further in defogging algorithms in basic step of image state recognition.