Results

The implementation of the project was a success.  The edge-detection algorithm that was used output a maximum of two frames per second without adaptive correction, and one frame per second in the adaptive case.

This performance was possible on the Xybernaut wearable computer, running at 233 MHz with a fast parallel port (for more information, see the Appendix).

As can be seen from the pictures in the Appendix, the edge detection does enhance the blurred image so that objects can be discerned.   The partially sighted person will still have blurred vision, but the edges will be blurred instead of an amorphous blob.

Applications

The obvious application for this project is to enhance vision for the partially sighted.  When wearable computing is cost-effective enough, and projects such as this one are refined and improved, it will be possible for ‘legally’ blind people to use their remaining vision in a functional context.

If the system were adapted to other image processing algorithms, it could be used for entertainment purposes in which modified video is an asset.   It could also be used for academic purposes as a real-time teaching tool for video and image processing.

Experiments in mediated reality are also possible with the edge-detecting system.  It would be interesting to see whether the outlines of objects create an attention focus to which the person’s visual field is drawn, much like the successful experiments in ASCII art.

Further Work

There are several improvements that can be made on the design. They are:

·        Implementing the image capture and edge detection on a hardware chip, especially an FPGA for quick reprogramming.

·        The project could be combined with the Linux Media Labs video capture card which operates at 30 frames per second.

·        The project could be integrated with the Linaccess project to make a complete open source system for the blind.