Simultaneous measurement of particle position and displacement in depth by depth-from-defocus using time-series color images

Abstract

This paper presents a method for simultaneous measurement of particle position and displacement in depth by depth-from-defocus using time series color images. This method enables us to obtain the particle information for 3D-PIV by using a single color camera. To improve the depth measurement accuracy, three differently blurred images are simultaneously captured as RGB color images and the depth information is estimated from the three images with a neural network considering in-plane position. In performance tests, we numerically and experimentally evaluate the measurement accuracy of particle position and displacement in depth. The results of depth position by neural network show that the RMS error decreases to 65% of that by a calibration line. The results of depth displacement from three color images show that the RMS error decreases to 24% of that from only one color image, corresponding to 9.5% for maximum displacement 2 mm.