This diploma thesis describes visualisation of large sets of measurement time-based vector data. The data is transformed from time to frequency domain so it can be analysed and detect various phenomena.
The problem we are facing is how to display all of the data at once. The goal is to reduce the size of the transformed data for a fast visualization while preserving its properties.
We reduce the data with data aggregation. The properties of the transformed data are preserved the most by using the aggregation with maximum value. With the described aggregation method a multi level circular buffer is designed, which reduces the data in both dimensions. Each next level of our buffer can display a longer time interval of reduced data than the previous one. The visualisation switches the displayed buffer level when the length of the time interval changes.
The visualisation is implemented with DirectX 11, where all transformations are done in shaders. It supports different views, data manipulation and basic user interactions like changing the time interval, rotation and translation of camera view and reading the value under the cursor.
|
