A Sharpness Measure for Image Quality Assessment Using Average Effective Number of Neighbors

The proliferation of portable and miniaturized imaging devices, coupled with the prevalence of communication networks have changed the way we create and share photos. Indices for image quality have been proposed extensively to evaluate the recorded photograph. In this paper, we first delineate the desirable properties of an image quality metric. We then describe a computationally effective approach to assess the sharpness of a photo so that images of poor focus can be identified. The proposed method attempts to measure the integrity of major structures by computing the effective number of neighbors (ENN) for strong edge pixels in an image. Simulations and experimental results indicate that this ENN-based metric conforms to all the desired properties of a quality metric and is able to estimate the blurredness effectively and efficiently.