Area Filling Algorithm Analysis And Improvement Of Its Application In The Snow Image Restoration

When it is snowing, people will go out to enjoy the snow scape, play snowball fights, make a snowman. The only way to keep the good moment is to photo them; Some people need to get images in the snow because of work. However, flying snowflakes would reduce the quality of images. Therefore, snow image restoration is particularly important. So, it is also one of the main jobs to promote the image restoration.In this paper, the image acquired under snowy conditions will be looked as the research object. By studying its characteristics and display characteristics in the image ranging from the size of the de-noising and snow patches fix main image contains. Snowflake is random in each image, and the size and shapes of snowflakes are varied. According to snowflakes image rendering mode, it is seen as small snow salt and pepper noise, and chunks of snow could be regarded as plaques. By analyzing and comparing several traditional de-noising methods, we selected the median filtering method in small snowflake image filtering process, in order to achieve the purpose of the removal of small pieces of snow. For bulk snow patches, we use the area filling methods Processing.Common seed filling algorithm and scanning lines filling algorithm are using pre-selected color, which would bring "scar", we improved scan line seed filling algorithm. First, selecting repaired area to the point as the center to the edges draw a number of straight lines. Then, calculate these lines with the intersection area to repaired borders, using color information on these boundary points neighbors domain points, obtaining an integrated border points color fill color value as a boundary point. From inside to outside the loop processing until all points is to be assigned to the repaired area completely with corresponding color. The algorithm does not require the use of pre-selected color filling, but to take full advantage of the area repaired in the border with similar color information. After that, image restored is closer to the real image. Finally, with the help of the Matlab and VC, we realize that the snowflake image repairing process works. Experimental results show that our method can be used to effectively deal with snowflake image, eliminate the influence of snow to the quality of the images.