On the reduction of cloud influence in natural and aerial images

Due to global warming, an increase amount of water could be found in the environment. It causes the appearance of clouds in satellite and aerial image acquisitions. Clouds and shadows cause interference in aerial images acquired, that can be in two ways: signal attenuation and different range of illuminations. The signal attenuation is caused by the image acquisition above the cloud cover, as in the case of ground image taken from airplanes or satellites. The cloud shadows on the ground can produce dark regions next to areas with direct sun light, making it difficult to identify equal textures on segmentation algorithms due to diverse amounts of illumination. This paper describes an algorithm to reduce both effects of clouds, workings in two layers. The outer layer uses a homomorphic system for multiplication. The inner layer is adaptive depending on the local contrast and average luminance. That is, it enhances image parts considering the degree of clouds in portions of the scene. The adaptive layer increases the high frequencies and reduces the low ones simultaneously. Moreover, depending on the conditions of acquisition, each layer could be used as an independent and individual system.