The vast majority of people today have reasonable vision.  They use their eyes to see things around them so they can interact with their surroundings.  However, partially sighted people who are considered 'legally' blind have extremely blurry vision that at most lets them see areas of light and darkness.

The visual enhancement system presented here uses wearable computing and image processing to enable partially sighted people to use their vision effectively.  Live video is captured using a digital camera and processed with a computer that is worn on the body.  The resulting video is displayed directly to the user's eye (with the EyeTap invention), creating a 'mediated reality' that can enhance vision.

The raw video is analyzed mathematically to perform an imaging technique called edge detection.  This process discovers high-contrast areas in a digital picture and outputs an map of these areas - the outlines of objects in the scene - in black and white.

The edge detection algorithm operates by differentiating the input image in the x and y directions using a 2-dimensional convolutional kernel, and combining the resulting images pixel-by-pixel with a Pythagorean operator.

Several different variants on the edge detection exist.  The one chosen for the project is called the Prewitt operator, which has speed and texture advantages over the Sobel and Canny operators.  The algorithm was further optimized through the use of lookup tables, integerization, and adaptive dynamic range correction.