Computing Local Space Average Color on a Grid of Processing Elements

2 Computing Local Space Average Color on a Grid of Processing Elements 

Each element has access to the color stored at position (x, y). We assume that we have three different color bands and that the color at position (x, y) is 

The computations are performed for all three color bands independently. Let the current 

We can initialize a, with i e r, g, b] with an arbitrary value. The following updates are then iterated indefinitely for all three color bands  

This percentage depends on the size of the image. For our experiments, we usually set this value to p = 0.0005 for images of size 256 x 256. In the subsequent text, we discuss how this value may be chosen. The first step averages the data obtained from neighboring elements. The second step fades the color c at position (x, y) into the current average a. This can be viewed as a biased average, e.g. 

---