基於智慧眼鏡之擴增實境輔助系統
摘要
近年來,擴增實境蓬勃發展,隨著智慧眼鏡的問世,更是讓這項技術為生活帶來許多的便利性。本論文提出一套擴增實境輔助系統,將系統架設於智慧眼鏡上,讓使用者得以透過個人視角使用此系統。 本系統功能包含(1)使用一套簡易流程建立裝置物件資料集,採用Mask R-CNN方法辨識裝置上的物件類別與位置(2)從影像中擷取指向手勢,並分析指向,依照手指指向顯示物件資訊(3)使用校準物件分析虛擬輔助工具該顯示的角度。 本系統的研發目的在於提供一套可輔助技術人員訓練之系統,藉由指向手勢,即可顯示操作人員欲了解的物件資訊。根據系統的實驗顯示,物件偵測的辨識率達到95.5%;手勢偵測的Kappa值為0.93,且平均偵測到手勢的秒數為0.26秒,即使在不同光線下,指向分析的準確率也有79%,由此可證明本論文所使用的方法具有很好的可信度。

關鍵字: Kinect、視覺障礙者、卷積神經網路、物品偵測

 

 

A Smart-Glasses-based Augmented Reality Assisted System
Abstract
Recently, the application of augmented reality has become more and more prevalent. With the advent of smart-glasses, the related research of augmented reality has been grown vigorously. Therefore, this dissertation proposes an augmented reality assisted system which will be set up on the smart-glasses. By setting the system on the smart glasses, the user will be able to use the system in personal perspective. There are three main features in the system.(1)With a set of simple procedure, the system will set up dataset of objects on the device. Moreover, the system can identify the object and its position on the device automatically by using the Mask R-CNN method.(2)By capturing pointing gesture from the image and analyzing the pointing direction, the system will display the object information according to the finger pointing direction.(3)Using the calibrating object to analyze the rotation angles of virtual tools. The aim of this system is to provide a system that can assist technicians in training. With the finger pointing, the system can show the object information which the user wants to know on the smart-glasses immediately. According to the results of the experiments, the percentage of recognition of object detection is 95.5%, the Kappa value of recognition of gesture detection is 0.93, and the average time for detecting pointing gesture is 0.26 seconds. Furthermore, even under different light, the proportion of accuracy of the pointing analysis is up to 79%. Based on the results of the experiments, it was proved that the method which was applied in this dissertation is applicable. Keywords: augmented reality, hand gesture recognition, finger-pointing analysis