CSE 477 -- Video Imaged Spatial Positioning Project |
|
|
Proposal |
Design IssuesWe have identified four main design issues. Precision: Embedded systems are only capable of integer operations. Because of this, we cannot attempt arithmetic operations that require decimal precision. Our initial research has shown that the coordinates we can produce will be within an inch of the real coordinates every 100 feet away from the cameras. Geometric Calculations: Because we cannot do floating-point arithmetic, we found two alternative methods: use a set of lookup tables or emulate floating-point arithmetic. Currently, we plan to use the table approach. Specifically, we need a table for pixel to angle translation and a table for a pair of angles to distance calculations. The table approach is feasible because each table will consist of a bounded recordset. But, at this point we are not sure of what the tradeoffs are between using tables and using floating-point emulations. Identifying the Marker in the Digital Image: In most cases, this will not be an issue. We are currently looking at identifying the marker through its color from the digital image. Yet to do this, we need to choose a marker color that will not blend into the surrounding environment. We are confident we can choose a marker color that will be distinct enough for most environments, but we know that we will not be able to find a universal marker color. This limits where our system can and cannot be used. We, currently, do not know the extent of the limitations. RF or Infrared: At this point, we are planning to use RF as the media to communicate between the XSV-300 circuit board and the user's PDA. Currently, we have not fully explored the tradeoffs between these two communication protocols. |