Gesture Recognition Algorithm for Visually Blind Touch Interaction Optimization Using Crow Search Method

Abstract

Touch screen interaction system is highly demanding for new innovations such as visual sensing, virtual key board system on the screen, three dimensional gesture communications methods, and RFID sensing etc. In spite of the existence of these types of interaction methods, visually impaired people struggle to have easy access to touch screens. The main goal of the research is to overcome the navigation problems that blind people face while interacting with touch screens. In this examination we centered to build up a Braille sketch; a motion put together information technique with respect to contact screens of advanced mobile phones for outwardly debilitated individuals. Utilizing Braille codes to perform motions on contact screen makes the outwardly tested people agreeable in light of the fact that “Braille is the reason for correspondence”. The streamlining procedure is the demonstration of amplifying or limiting a genuine function by systematic choosing input parameters from an accessible pool of parameters and to figure the estimation of function. Here, we accept variables as hand finger motion facilitate, for example, the coordinate values on x and y axises, swipe limit speed, swipe least separation, pixel rate and speed of X and Y. To build the execution of the system structure, we improve by differing shrouded layer and neuron using Crow Search Algorithm (CSO). The ANN with CSA achieves the Optimal Hidden Layer and Neuron (OHLN) to anticipate the right motion yields. These strategies present an answer that will consequently perceive the hand signals so impeded individual can without much of a stretch speak with ordinary individuals. The proposed model will give high precision with ideal execution measurements contrasted with other existing created demonstrate.

Keywords

• Hand gesture
• ANN
• Braille
• hidden layer and neuron
• blind people mobile interaction
• machine learning algorithms