An object recall system based on 3D vision techniques

• Main Authors: Chu-Cheng Huang, 黃竹政
• Other Authors: Chin-Pan Huang
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/36430369618181704878

碩士 === 銘傳大學 === 資訊傳播工程學系碩士班 === 103 === This study presents a 3D vision based system which is used to assist users for recalling household objects placement to enhance the convenience of the people in their daily lives. To achieve the aim of the system, we propose to integrate three procedures of tracking, segmentation, and recognition. First, utilize the depth camera (Kinect) to track the user''s hands and apply the skin detect scheme to determine whether the user is fetching object. Then employ the 3D visual information of the depth camera to segment and position the object and remove the unwanted information at the same time. Finally, use the segmented test object to match the reference household objects in the database to achieve recognition purpose. Experimental results have demonstrated effectiveness and promising performance of the proposed method.