Date of Publication :7th April 2016

Abstract: In the present era of smart electronics, hand sign recognition has a significant role. A hand sign recognition algorithm based system employing a hybrid SOM-Hebb network is designed. The input posture image is pre-processed and feature vectors are extracted from them. These are then mapped on to lower dimensional neurons on the self-organizing map (SOM). The input image is compared with the hand posture images in the database. The algorithm offers a stagnant response irrespective of the change in location of the input image. The work is to develop a design that supports real- time FPGA implementation using onboard camera. The set of default images in database used as the current input. The input image is pre-processed and converted to a binary image. The feature vectors are extracted using Discrete Fourier Transform (DFT) and fed to the hybrid classifier network for recognition. FPGA implementation enhances its use in portable embedded applications. The system offers better recognition accuracy and faster response. The system is designed for robustness against change in location of input image. The algorithm is coded using VHDL and simulated in Xilinx ISE 13.2 and MATLAB 2013.

Reference :