3D positioning and autofocus of the particle field based on the depth-from-defocus method and the deep networks

Abstract Accurate three-dimensional positioning of particles is a critical task in microscopic particle research, with one the main challenges being measurement depths. In this paper, we propose method for detecting depths from their blurred images using depth-from-defocus technique and deep neural network-based object detection framework called you-only-look-once. Our provides simultaneous lateral position information has been tested evaluated on various samples, including synthetic particles, polystyrene blood cells, plankton, even noise-filled environment. We achieved autofocus target different generative adversarial networks, obtaining clear-focused images. algorithm can process single multi-target image 0.008 s, allowing real-time application. proposed new opportunities field research.