Adaptive contrast enhancement involving CNN-based processing for foggy weather conditions & non-uniform lighting conditions

Adaptive image processing in the context of Advanced Driver Assistance Systems (ADAS) is a crucial issue because bad weather conditions lead to poor vision. In a foggy weather, image contrast and visibility are low due to the presence of airlight that is generated by scattering light, which in turn is caused by fog particles. Since vision based ADAS are affected by inadequate contrast, a real-time capable solution is required. To improve such degraded images, a method is required which processes each image region separately. Hence, real-time processing is required, the method is realized with the CNN paradigm which claims the characteristic of real-time image processing. To compare the proposed method with existing state-of-the-art methods the Tenengrad measure is applied.