Evaluating sharpness metrics for HE stained blood cell images

Image sharpness is a feature that has great influence on image quality. In this paper we present methods to evaluate the sharpness of images, that can be used to select the best image in a sequence of images taken of the same scene using different focus settings. We then execute measurements on 1195 image sets using two sharpness metrics - one based on image gradient sums and one based on the edges found in the images. The metrics are evaluated on original noise free images, different amount of noise is added to the images and the metrics are reevaluated in those circumstances as well. Two additional evaluation properties are proposed in this article: roboustness and position of the sharpest image in a miss.