A vision-based technique for assisting visually impaired people and autonomous agents

The paper overviews a vision-based technique that can be instrumental in assisting visually impaired humans in their activities. This category includes both blind people and individuals unable (e.g. due to neural disabilities) to understand semantics of the perceived visual data. Alternatively, the technique can be used as a tool for autonomous agents (e.g. mobile robots) performing in complex environments. No model of the world is used, and the technique is equally suitable to both indoor and outdoor applications. From the machine vision perspective, the method consists in detecting nearly identical (i.e. photometric/geometric distortions and partial occlusions are allowed) fragments in images of unknown and unpredictable contents. The paper focuses on the machine vision aspects of the problem (theory, exemplary results, computational efficiency, etc.). The future development of actual systems assisting the handicapped is only preliminarily highlighted.