The Illumination Analysis Of Haze Outdoor Scene With Basis Images

Augmented Reality is more and more popular in computer science research recently years, which integrates virtual objects generated by computer into the real scene and make the two parts looks harmonious to be a same scene. AR technology is widely used in such many areas as TV and film industry, interior design, city planning, military simulation training and so on. In order to achieve the seamless fusion of virtual and reality, maintaining consistency between virtual objects and the real scene is very Significant. However, illumination consistency is one of them.Images captured in outdoor scene are effected by many facts in nature world, such as fog, haze, rain and cloud. Because of environment pollution, we are easier to get a hazy weather in almost every city nowadays. In hazy weather condition, images of outdoor scenes are degraded by the strong scattering of atmospheric particles. Before we get the right illumination analysis of hazy images of outdoor scene, we need to do haze removal first.In hazy weather condition, atmospheric particles are larger than usual. These particles have strong effect on light scattering. On the one hand, the irradiance received by the camera from the scene point is attenuated along the line of sight. Brightness is reduced. Furthermore, ambient light reflected into the line of sight (we call it airlight). So, the atmospheric transmission model in hazy weather composed by these two parts.In order to achieve the right illumination analysis, we analyze the hazy images combined with hazy weather model and the basic images of the same scene. We got the depth of the scene points when we got the haze-free images. Besides, we can got the right scattering coefficient of the atmosphere. In this paper:1. We decompose an image of outdoor scene into the linear combination of the skylight basis image and the sunlight basis image. The combination coefficients are the intensity of the skylight and sunlight.2. We initialized some scene points and the beginning scattering coefficient (hazy density). Combined with the hazy weather model and basis images, we calculate the rough depth image of the whole scene points. 3. In our theory, the haze-free image and combination image should be one. We minimum the hue difference of these two images, then we got the proper hazy density (scattering coefficient).4. After we got the proper haze-free image, we re-calculate the intensity of the sunlight and skylight. And we put the corrected intensity of sunlight and skylight into the cycle of minimum the hue difference of haze-free image and combination image,then we got the corrected depth image, hazy density and the haze-free image.