CSE 410 - Project 3 Images

University of Washington Department of Computer Science & Engineering

CSE Home

About Us

Contact Info

Here are some images created with the solution code for project 3. Note that the images on this page are all gif format for ease of display. The project 3 program can only read 24-bit bmp format files. The original bmp formatted test images ( rgb200.bmp and tn-fish.bmp ) are available if you want to do similar experiments.

A couple of initial images were provided.

rgb200: basic image for testing the filters tn-fish: another small image for testing the filters

The first filters are flip and mirror. Since the rows are all word-aligned, the flip filter can do word copies. The mirror filter does byte by byte copies since the pixels are not word aligned.

rgb200-1: flip rgb200-1-2: flip, then mirror

There are filters to brighten and darken the various color planes. In both of these images, the red is brightened. The color values must be clamped so that they don't exceed 255 and wrap around to 0, or conversely, go below 0 and wrap around to 255.

rgb200-17-17-17-17-17-17: brighten red tn-fish-17-17-17-17-17-17: brighten red

There are some simple filters to select one or more of the color planes.

rgb200-11: select red and green tn-fish-10: select green

The threshold filters can create interesting effects. ="tn-fish-33.gif"

tn-fish-33-2: threshold red, mirror tn-fish-33: threshold red

For extra credit, one could also do simple convolution filters like mean blur, and the Laplacian of the Gaussian. In the solution, these 5x5 convolution procedures pretty much filled every available register.

tn-fish-3: blur rgb200-4: Laplacian of Gaussian

Department of Computer Science & Engineering
University of Washington
Box 352350
Seattle, WA 98195-2350
(206) 543-1695 voice, (206) 543-2969 FAX
[comments to webmaster]


	A couple of initial images were provided.
rgb200: basic image for testing the filters		tn-fish: another small image for testing the filters

		The first filters are flip and mirror. Since the rows are all word-aligned, the flip filter can do word copies. The mirror filter does byte by byte copies since the pixels are not word aligned.
rgb200-1: flip	rgb200-1-2: flip, then mirror

There are filters to brighten and darken the various color planes. In both of these images, the red is brightened. The color values must be clamped so that they don't exceed 255 and wrap around to 0, or conversely, go below 0 and wrap around to 255.
	rgb200-17-17-17-17-17-17: brighten red	tn-fish-17-17-17-17-17-17: brighten red

		There are some simple filters to select one or more of the color planes.
rgb200-11: select red and green	tn-fish-10: select green

	The threshold filters can create interesting effects.
tn-fish-33-2: threshold red, mirror		tn-fish-33: threshold red

For extra credit, one could also do simple convolution filters like mean blur, and the Laplacian of the Gaussian. In the solution, these 5x5 convolution procedures pretty much filled every available register.
	tn-fish-3: blur	rgb200-4: Laplacian of Gaussian