scispace - formally typeset
Search or ask a question

Showing papers by "Chien-Hung Liu published in 2015"


Journal ArticleDOI
TL;DR: The methods proposed in this study can be used in public safety, student attendance registration, commercial VIP recognition and many others, and time spent searching and comparing characters is reduced by creating height categories.
Abstract: A great deal of relevant research on character recognition has been carried out, but a certain amount of time is needed to compare faces from a large database. The Kinect is able to obtain three-dimensional coordinates for an object (x & y axes and depth), and in recent years research on its applications has expanded from use in gaming to that of image measurement. This study uses Kinect skeleton information to conduct body height measurements with the aim of improving character recognition performance. Time spent searching and comparing characters is reduced by creating height categories. The margin of error for height used in this investigation was ± 5 cm; therefore, face comparisons were only executed for people in the database within ±5 cm of the body height measured, reducing the search time needed. In addition, using height and facial features simultaneously to conduct character recognition can also reduce the frequency of mistaken recognition. The Kinect was placed on a rotary stage and the position of the head on the body frame was used to conduct body tracking. Body tracking can be used to reduce image distortion caused by the lens of the Kinect. EmguCV was used for image processing and character recognition. The methods proposed in this study can be used in public safety, student attendance registration, commercial VIP recognition and many others.

16 citations


Journal ArticleDOI
TL;DR: In this paper, an optical sensor for the real-time measurement of axial thermal elongation of a machine tool high-speed spindle is described. But the accuracy of this optical sensor is better than 1μm within the measuring range of ±100μm and the triple standard deviation is 0.03μm.
Abstract: This paper describes an optical sensor for the real-time measurement of axial thermal elongation of a machine tool high-speed spindle. A cat's eye optical reflector embedded in the spindle reflects the image of a laser diode beam onto a quadrant photodiode detector. Changes in the position of the reflected laser spot on the photodiode in a detection module allows determination of changes in length of the spindle. The resolution is 50 nm for the low frequency range. The accuracy of this optical sensor is better than 1 μm within the measuring range of ±100 μm and the triple standard deviation is 0.03 μm. The experimental results showed that thermal elongation can be measured and controlled to within 2 μm of elongation and shortening variation within the stated range using an oil cooling system.

8 citations