Abstract
Hand gestures are natural means of communication for human beings and even more so for hearing and speech impaired people who communicate through sign language. Unfortunately, most people are not familiar with sign language and an interpreter is required to translate dialogues. Hence, there is a need to develop a low cost, easily implementable and efficient means to recognize sign language gestures to eliminate the interpreter and facilitate easier communication. The proposed work achieves a satisfactory recognition accuracy using in-built laptop webcam using combination of 3 skin color models(HSV,RGB,YCbCr) and background subtraction to eliminate noise from webcam low quality images to recognize sign language for helping the hearing and speech impaired in real-time without requiring too much computational power or any other device as it can be implemented in any laptop with a webcam.
Highlights
Hand Gesture Recognition system is an active research area in computer vision due to its broad application in human computer interaction(HCI),sign language, augmented reality,gaming,etc
Problem Formulation This paper proposes a method to recognize static hand gestures in complex background images that represent alphabets and numbers of the standard Polish Sign Language
The training dataset used to train the classifier consists of a total of 100 images of which 50 images are taken by the DSLR camera with a good resolution and uniform background, and other 50 images are taken using a webcam
Summary
Hand Gesture Recognition system is an active research area in computer vision due to its broad application in human computer interaction(HCI),sign language, augmented reality,gaming,etc. Static gestures are expressed in single frames whereas dynamic gestures span over multiple frames incorporating temporal variability To recognize these gestures there are three basic processing stages- hand segmentation, feature extraction and classification. Hand segmentation contributes greatly to the accuracy of gesture recognition which poses many difficulties because hand segmentation is affected by illumination changes, skin color differences between humans, presence of other background elements having similar skin color and low camera quality. To solve this issue of effective hand segmentation many methods have been proposed. The paper is organized as: Section A describes the image preprocessing, Section B illustrates the training stage and Section C illustrates the testing stage
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
