Comparative analysis of autofocus measures for blood film microscopy

Abstract

This study compares 28 focus metrics that are utilized for computer microscopy autofocusing, specifically for blood film analysis. Three datasets are evaluated in the study: defocused image stacks, blurred blood film images, and synthetic sinusoidal images. Every image set is subjected to focus measure computations using 28 metrics, including histogram, statistical, and derivative-based approaches. Accuracy, Range, False Maxima, Width at 50% Maximum, and Overall Score are the five main variables used to evaluate the performance of these focus measures. The findings show that derivative-based metrics, like Squared Gradient and Variance of Gradient (VoG), fared far better, with excellent accuracy, a wide focus range, and less vulnerability to false maxima. On the other hand, metrics like Threshold Absolute Gradient that had larger peaks, and high False Maxima were less dependable for accurate autofocusing. This study demonstrates how well derivative-based focus measures work to achieve accurate and dependable focus detection in applications involving microscopy.