Research of Snow-Degraded Image Clearness Methods
Kai Song, Dong Hun Yang, Xin Nie · Applied Mechanics and Materials · 2013
An improvement of coterminous frames differencing is proposed. By using the improved algorithm, the snow point from video image obtained in the snow day can be removed significantly. The snow regions are extracted by the temporal difference of pixels from the five adjacent frames in video images, The absolute value of differential luminance of each frame is then obtained. The luminous intensity of the background and effects by snow can be further computed. Finally, the intensity of the contaminated pixels is replaced by the average value of future and past frames. The results of the experiment show that this algorithm can improve the quality of video images in light snow, heavy snow even heavy snowstorm strongly.