期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2014
卷号：69
期号：1
出版社：Journal of Theoretical and Applied
摘要：The advent of modern sciences has played a pivotal role in exploring other, if not better, avenues like the vision based hand interface techniques instead of the more traditional means of those of keyboard, mouse, pen etc, wherein the former has attained a higher relevance because of its natural and efficient outcomes. Through this paper an attempt has been made at developing an algorithm capable of intercepting the hand moments and locating its centroid. The proposed method tabulates the corresponding feature attributed to a gesture, after analyzing and intercepting the distance between the finger tips, their count, relative distance between the finger tips and their distance from the centroid, which eventually adds up to creation of the histogram of an image through the substantial and collateral information received through the hand gestures. The proposed technique works against a yellow background with interception of gestures of the front and the back side of hand. In an attempt of enhancing the robustness of the processing in consideration of the environmental impact through noise, the HSV format is taken into use. This method works on intercepting dedicated gestures of fingers to identify different alphanumeric characters followed by tracking of hand using motion and color cues. We have investigated the application of the method in sign language recognition by dedicated hand gestures feature learning. Through the instant method a proposal to an approach for recognition of sign language having usage in machine learning problem is being made which undoubtedly would be extremely helpful for deaf people to interact with others who are unaware of or don�t understand Sign Language.
关键词：Machine Learning; American Sign Language (ASL); Skin Detection; Gesture Recognition; Image Tracking; Histogram.
