Mobility for visually impaired people in an outdoor environment has always been a challenge. Traditionally, white cane or guide dogs have been used to help in mobility, but they suffer from many limitations. Modern solutions like Smartcane or Ultracane can help detect obstacles in the path, but cannot differentiate among these obstacles. Moreover, they do not support navigation. Considering the limitations with existing aids for mobility of visually impaired, we propose a new device to help visually impaired in their outdoor mobility. This device uses an RGB camera and image processing techniques to recognize the surrounding objects and inform/alarm the user accordingly. Our prototype incorporates signboard detection, face detection and surface texture detection to address issues related to mobility of a visually impaired user. Localization module implemented using GPS and IMU helps to support navigation. In this paper, we explain our initial implementation of various image processing tasks. We focus on the exploration for the choice of algorithms, design decisions, and various implementation decisions. We explain the challenges and results of implementing these applications on an embedded platform (Zedboard) for prototyping. © 2016 IEEE.