1.2.2. Progressive Vector Transmission

Progressive vector transmission deals with the problem of viewing large volume vector datasets by progressively adding data to a model with less detail. On the one hand, this means data can be streamed to a client rather than downloaded in a single chunk. On the other, it means that the amount of detail shown can be limited to the scale at at which it is being shown and the resolution of the screen on which it is being displayed.

Steps

Progressive vector transmission has been studied by a number of different researchers for example, (Bertolotto et al. 2001), (Buttenfield 2002), (Brenner et al. 2004), and (Yang 2005). All these approaches share the same basic steps. We will draw on the approach of (Yang 2005) to illustrate these. This streams data for lines and polygons to a client.

The first step answers the question of how to select a meaningful portion of the data. The method used here employs the line simplification algorithm of (Visvalingam et al. 1993). This allows points of a line to be ranked in order of their importance to the shape.

The second step is how to organise the data. Since the points have been ranked they can be sent to the client in order of importance. To do this a set of instructions is generated that tells the client what to do with a point when it recieves it.

The third step is to decide how much data to show. This is not an easy question to answer. One method in cartography is to use the Radical law ( (Toepfer et al. 1966), (Dutton 1999)), which states the total number of points that should be shown at different scales. The equation is shown below.
In the equation: n_f is the number of objects that should be shown on the map at the final scale. n_a is the number of objects at the source scale. M_a and M_f are the denominators of the source and derived scales respectively. The exponent X is a parameter that can be varied according to the feature type. Toepfer and Pillewizer specified that a value of 1 should be used for point symbols, 2 for linear symbols (including points along digitized boundaries) and 3 for areal symbols (e.g. islands).

Example

Use the animation to explore a data set that has been organised for progressive transmission. You will need to select one of two modes. In the transmission mode the scale is constant, but you can vary the number of points displayed by moving the slider. In the zoom mode you can change the scale of the dataset which will cause the data to be zoomed and the amount of detail to be varied. You can also change the exponent value used in the Radical law.

Experiment with the choice of exponent in the Zoom view. What do you think is a good exponent for the Radical Law? How does your choice compare to the suggestions of Topfer and Pillewizer? If it is different, Why should it be different? (Click here for more information)