Advanced Image Processing for Computer Vision Systems for Visually Impaired Navigation

Assistive technology enables people to achieve independence in the accomplishment of their daily tasks and enhance their quality of life. Navigation systems to assist visually impaired people that use computer vision techniques are very resource intensive and as such, not suitable for mobile devices. This work describes a computer vision algorithm intended to reduce processing demands with possible application in portable devices. The model proposed employs 2D-Ensemble Empirical Mode Decomposition (2D-EEMD) in order to reduce the problem complexity. It uses a combination of Peano-Hilbert space filling curves as a dimension reduction of image data and EEMD to implement a fast algorithm. Tests results have shown that Peano-Hilbert EEMD exhibits substantially reduced processing power compared with 2D-EEMD approaches, enabling new opportunities for its use in portable applications on low computer power devices.