Dehazing of Natural Images using Non-Linear Wavelet Filter

The outdoor images are captured in bad weather conditions to yield poor visibility conditions, where noise is present due to low contrast of an image, color differentiation which is a major problem for most computer vision applications. Haze is caused when an image is captured by the camera between the image is strike with the haze because of attenuation and air-light. Attenuation reduces the contrast level of the image and air-light increases the brightness of the image. Haze removal techniques recover back the contrast of the scene. The existing de-hazing method on an atmospheric scattering model and therefore it has the common limitation that is this model is valid only when the atmosphere is homogeneous. This paper reveals to estimate the threshold value to reduce the noise in outdoor images by using wavelet thresholding method. By calculating the value of threshold and types of wavelet functions, this is an important issue in de-noising based wavelet approach. It is possible to perform non-linear de-noising by thresholding the wavelet coefficients. In this paper, the result shows the de-noising algorithm based on the wavelet de-hazing method in terms of increasing the visual quality of the image, this proposed algorithm not only exclude the noise but also improve the PSNR to get better image quality.