ForkLoop Images Assignment
Contents
Motivation
We gain experience using a parallel fork loop. Image processing often presents the opportunity for a lot of parallelism. Here we will operate on a single image, making a parallel task for each row.
Background
Revisit the Image Batch Race Condition Exercise's Background section for information on pixels and images.
Utilities
Images of Rows of Colors
For this assignment we will be dealing with images. For clarity, we have structured images as a list of rows. Each row is structured as a list of colors. We use the HSV representation of color.
interface ImmutableHsvImage and class DefaultImmutableHsvImage
interface HsvImage
interface ImmutableHsvImage extends HsvImage
class DefaultImmutableHsvImage implements ImmutableHsvImage
interface ImmutableHsvRow and class DefaultImmutableHsvRow
interface ImmutableHsvRow
class DefaultImmutableHsvRow implements ImmutableHsvRow
class HsvColor
class HsvColor
PixelFiler
interface PixelFilter extends UnaryOperator<HsvColor>
Code to Investigate
SequentialImageFilter
The provided SequentialImageFilter shows off how to read from a source image and construct a new (destination) image. The specified pixelFilter is used to convert the source pixel color into its destination color.
Note: the interfaces ImmutableHsvRow and ImmutableHsvImage are used when appropriate. When we need to construct instances of those interfaces, we use classes which implement them: DefaultImmutableHsvRow and DefaultImmutableHsvImage.
Note: the rowCount() and columnCount() methods are only used to provide the (not required) initialCapacity when constructing the ArrayLists.
| SequentialImageFilter |
|---|
public class SequentialImageFilter implements ImageFilter {
@Override
public ImmutableHsvImage apply(ImmutableHsvImage src, PixelFilter pixelFilter) {
List<ImmutableHsvRow> dstRows = new ArrayList<>(src.rowCount());
for (ImmutableHsvRow srcRow : src.rows()) {
List<HsvColor> dstRowPixels = new ArrayList<>(srcRow.columnCount());
for (HsvColor srcPixel : srcRow.columnColors()) {
HsvColor dstPixel = pixelFilter.apply(srcPixel);
dstRowPixels.add(dstPixel);
}
dstRows.add(new DefaultImmutableHsvRow(dstRowPixels));
}
return new DefaultImmutableHsvImage(dstRows);
}
}
|
DesaturateAllPixelFilter
| class DesaturateAllPixelFilter |
|---|
public class DesaturateAllPixelFilter implements PixelFilter {
@Override
public HsvColor apply(HsvColor src) {
return new HsvColor(src.hue(), 0.0, src.value());
}
}
|
Code To Investigate
PixelFilters
Our Image Filter will apply a PixelFilter to every pixel in a specified image.
Some example pixel filters have been provided which use named classes, anonymous classes, and lambdas:
| class PixelFilters |
|---|
public class PixelFilters {
private static class HueSettingPixelFilter implements PixelFilter {
private final double hue;
public HueSettingPixelFilter(double hue) {
this.hue = hue;
}
@Override
public HsvColor apply(HsvColor src) {
double saturation = src.saturation();
double value = src.value();
return new HsvColor(hue, saturation, value);
}
}
public static PixelFilter hueSettingPixelFilterOf(double hue) {
// return new instance of named class HueSettingPixelFilter
return new HueSettingPixelFilter(hue);
}
public static PixelFilter hueAdjustingPixelFilterOf(double deltaHue) {
// return new instance of anonymous class
return new PixelFilter() {
@Override
public HsvColor apply(HsvColor src) {
double hue = src.hue();
double saturation = src.saturation();
double value = src.value();
return new HsvColor(hue + deltaHue, saturation, value);
}
};
}
public static PixelFilter saturationAdjustingPixelFilterOf(double deltaSaturation) {
// return anonymous function (also known as a lambda) with parameter type
return (HsvColor src) -> {
double hue = src.hue();
double saturation = src.saturation();
double value = src.value();
return new HsvColor(hue, saturation + deltaSaturation, value);
};
}
public static PixelFilter valueAdjustingPixelFilterOf(double deltaValue) {
// return anonymous function (also known as a lambda) without parameter type
return src -> {
double hue = src.hue();
double saturation = src.saturation();
double value = src.value();
return new HsvColor(hue, saturation, value + deltaValue);
};
}
}
|
Code To Implement
ParallelTaskPerRowImageFilter
| class: | ParallelTaskPerRowImageFilter.java |  |
| methods: | apply | |
| package: | imagefilter.exercise | |
| source folder: | student/src/main/java |
apply
method: public ImmutableHsvImage apply(ImmutableHsvImage src, PixelFilter pixelFilter) 
(parallel implementation required)
Apply the provided pixelFilter to each pixel in the src HsvImage.
Each pixel is represented by an HsvColor.
PixelFilter has one interesting method apply which takes a source HsvColor and returns a new filtered HsvColor. It will NOT modify the passed source HsvColor. It would be impossible anyway, as HsvColor is immutable.
Example use of PixelFilter apply:
HsvColor dstPixel = pixelFilter.apply(srcPixel);
To complete this method we must return a new instance of the HsvImage class.
Nicely, HsvImage provides both:
- a constructor which takes an List<HsvColor[]> and,
- another constructor which takes an HsvColor[][].
These two constructors match up well with the two different return types of the many incantations of join_fork_loop:
where R is the return type of the TaskFunction<T, R> passed to the join_fork_loop.
What should the return type from the this TaskFunction be?
| Requirements: |
 Create a task per row. |
| Do NOT create a task per pixel. |
| Use a parallel fork loop. |
Visualization
| class: | ImageFilterApp.java | VIZ |
| package: | imagefilter.viz | |
| source folder: | student/src/main/java |
Testing Your Solution
| class: | __ParallelTaskPerRowImageFilterTestSuite.java |  |
| package: | imagefilter.exercise | |
| source folder: | testing/src/test/java |
Pledge, Acknowledgments, Citations
| file: | exercise-fork-loop-images-pledge-acknowledgments-citations.txt |
More info about the Honor Pledge
