National ICT Australia has developed AutoMap, a Linux-based system designed to make personal navigation systems more accurate. AutoMap uses machine-vision techniques to detect and classify geometric shapes from video footage, including shapes such as signs and company logos, which can change frequently in a neighborhood and make it difficult for digital map makers to keep their projects up to date.
AutoMap project leader Lars Petersson says an average of 10%–15% of street signs in an area will change every year. He says mapping companies currently employ someone to drive up and down each street in a van equipped with five or six cameras, with a passenger making annotations. Collected footage is then examined frame by frame to determine the location of signs.
Petersson says this manual task frequently results in numerous mistakes. The AutoMap system automatically detects signs from video footage, and with further research Petersson expects to have developed more advanced data-gathering techniques. One possibility is placing a small camera inside taxis, fleet vehicles, and garbage trucks.
"These vehicles will traverse the whole road network on a regular basis," Petersson says. "They will be able to automatically detect points of interest and automatically send this information back to base where a complete and constantly updating map emerges over time." The researchers also plan to develop methods capable of recognizing three-dimensional images such as park benches and speed cameras.
From Computerworld Australia
Abstracts Copyright © 2009 Information Inc., Bethesda, Maryland, USA
No entries found