Combining Feature Selection with Neural Networks for Polish Sign Alphabet Recognition

Nowadays there are many researches on sign language recognition, but just a few with regards to Polish Sign Language. It is not a trivial topic, because there is still a barrier between the Hearing and Deaf people. Different techniques such as machine learning can be applied to recognise gestures and to help to overcome the problem. The purpose of this work is to ease the communication between the Deaf and Hearing. During the work, we focused on the letters of the Polish Sign Language alphabet and digits. In the research, we proposed some new feature extraction algorithms. To classify signs the neural networks have been used. The experimental results confirmed the effectiveness of our solution.