Content-adaptive image enhancement, based on sky and grass segmentation

Current TV image enhancement functions employ globally controlled settings. A more flexible system can be achieved if the global control is extended to incorporate semantic-level image content information. In this paper, we present a system that extends existing TV image enhancement functions with control signals that are adaptive to grass and sky areas in the image. We investigate whether this content-adaptive control can offer higher levels of enhancement compared to the current global control approach. Our simulations on 117 natural images show that the mean square pixel value difference from optimally enhanced images reduces with 60%, when content-adaptive control is employed instead of a global controlled approach.