Palm-Finger Segmentation Using R-Theta With Scan-Line

• Main Authors: Shih-Han Chang, 張詩涵
• Other Authors: Shih-Shinh Huang
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/69051656832795525152

碩士 === 國立高雄第一科技大學 === 電腦與通訊工程研究所 === 99 === In this work, we propose a computer vision method for cutting wrist. The challegnes of this problem include complicated background and geometry variance, such as translation, rotation, and scaling. Accordingly, the main objective of our work is to effectively remove the wrist by taking the aforementioned challenges into consideration and thus facilitate the further gesture recognition. The gestures used in this work are from gesture zero to gesture nine. Firstly, we represent skin color distribution by a Gaussian model, and then use thresholding strategy to determine the hand region according to skin Gaussian model. For obtaining a more complete hand image, we apply Region Growing algorithm to the segmented region. Then, we use the R-Theta method for analyzing the hand region in order to find the wrist position. The purpose of using R-Theta is to achieve rotation invariance. Finally, the use of horizontal scanning method finds out the Wrist-Line between the wrist and palm, and then the wrist can be successfully removed for further consideration. In experiment, we validate our proposed method by using 1,600 images and achieve the cutting accuracy about 88.4%.