Information signs compression and classification using vector quantization and neural network for blind man tourisms navigation system

This paper presents the information signs compression and classification in vision-based robot guidance system that apply for blind man tourism navigation system which can have two main roles that first for image signs compression and next for signs classification. The algorithm described here take an advantage of image sign features that their colors and shapes are very different from natural environments. The system are divided into three parts, first for image detection and compression that are proposed the image encoding and decoding algorithms called wavelet transform with Fuzzy C-Means (WT+FC) via vector quantization techniques (VQ). The small bit rates for high-speed data transmission with a small space for data storage are required on Wi-Fi Channel. Simultaneously, the peak signal to noise ratio (PSNR) has to be maintained. The shape analysis with a continuous thinning algorithms and image binary data encoding algorithm are used in second part for reduced the sized of data and can be representatives for suitable features data to classify. Finally the back propagation Neural Network (BNN) techniques are used in image recognition and decision process and display the right task. By applying the proposed method, performance has been improved which indicated by lower bit rate and better PSNR. Some results from natural scenes are shown that system performance can work well and is valid to detect other kinds of signs that would train the mobile robot to perform some task at that place.