Resume Biography Gallery Digital Images Projects
Josh Grant
 
Master's Thesis
IVTrace
Creating 2D Images
Filter Functions
3D Elastic Collisions
3D Grapher
Open Inventor
Programming Tools
BoundingBox
ExaminerViewerPlus
MarchingCubes
ScalarArithmitic
SFScalarField
Courses
Photo-Realistic 2002
Visualization 2001
Photo-Realistic Computer Graphics
Intro to Computer Graphics
Scientific Visualization
 
Filter Functions
May 2001
Box Filter

The simplest of filters is the Box filter. If the distance of surrounding pixels is less than or equal to the radius of the dot, then set the color of that pixel to the center's color (give it equal weight). The graph to the left is an illustration of what the surrounding pixels weights are.

The image at left is an example of drawing a dot and applying a Box filter to it.

d = distance from center
r = radius of dot
if (d <= r) then
  pixel weight := 1;
else
  pixel weight := 0;


Tent Filter

The Tent filter is appropriately named if you take a look at the graph at left. The weight of each surrounding pixel changes linearly with distance from the center. This is a simple idea which produces far better results than the Box filter.

The image at left is an example of drawing a dot and applying a Tent filter to it. Notice the small bright center where the weight is equal to one.

d = distance from center
r = radius of dot
if (d <= r) then
  pixel weight := 1 - d/r;
else
  pixel weight := 0;


Quadratic Filter

Sometimes we don't want the surrounding pixels weights to drop off so quickly. To resolve this a Quadratic filter can be used. Now many of the surrounding pixels have weights closer to one before dropping off to zero (refer to the graph at left). The Quadratic filter produces dots which are brighter than the Tent filter and has softer edges than the Box filter.

The image at left is an example of drawing a dot and applying a Quadratic filter to it. Notice how it resembles a sphere.

d = distance from center
r = radius of dot
if (d <= r) then
  pixel weight := 1 - d2/r2;
else
  pixel weight := 0;


Cubic Filter

Another filter which produces results inbetween those of the Quadratic and Tent is the Cubic. In general it does not produce results far better than the Quadratic and is not used as much.

The image at left is an example of drawing a dot and applying a Cubic filter to it.

d = distance from center
r = radius of dot
if (d <= r) then
  pixel weight := 1 - 3d2/r2 + 2d3/r3;
else
  pixel weight := 0;


 

Josh Grant > Projects > Filter Functions

Comments or questions about this page can be addressed to Josh Grant at grant@cs.fsu.edu