Fast way to manipulate each pixel from ImagePlus - java

Hello I need to access every pixels of an ImagePlus for image analysis.
Because of the huge amount of images to process, I was wondering if there are special effective ways/methods to access and/or modify each pixel from an imagePlus?
The only idea I naturally come out with is double for-looping through the image matrix, which takes me several dozens of seconds to achieve on a 1000x1000 image.
Here is my code:
ImagePlus Iorg = IJ.openImage("Demo1.png");
int[] pix = Iorg.getPixel(5, 5);
if(Iorg.getSlice() != 1) {
System.exit(0);
}
for(int w=0; w< Iorg.getDimensions()[0]; w++) {
for(int h=0; h<Iorg.getDimensions()[1]; h++) {
System.out.println(w + " x " + h);
// DO what needs to be done
}
}
Any idea?

Since images are uchar, what you want to do is the equivalent of
if(selected_pixel==255)
selected_pixel = 1;
else
selected_pixel = 0
You can create mask, that would be easier. I don't know in java ImagePlus but in matlab it is mask = image==255;.
Try to use those kind of matrix operations according to your need. I'm sure these methods should be somewhere inside the library(if ImagePlus is image processing library.)

Related

Pixel relocation, showing side-by-side

I want to read individual pixels from one image and "relocate" them to another image. I basically want to simulate how it would be if I grabbed pixel by pixel from one image and "move" them to a blank canvas. Turning the pixels I grab from the original image white.
This is what I have right now, I'm able to read the pixels from the image and create a copy (which comes out saturated for some reason) of it.
import java.awt.Color;
import java.awt.image.BufferedImage;
import java.io.File;
import java.io.IOException;
import javax.imageio.ImageIO;
public class ImageTest
{
public static void main(String args[])throws IOException
{
//create buffered image object img
File oldImgFile = new File("/path/to/image/shrek4life.jpg");
BufferedImage oldImg = null;
BufferedImage newImg = null;
try{
oldImg = ImageIO.read(oldImgFile);
}catch(IOException e){}
newImg = new BufferedImage(oldImg.getWidth(), oldImg.getHeight(), BufferedImage.TYPE_INT_ARGB);
File f = null;
try
{
for(int i = 0; i < oldImg.getWidth(); i++){
for(int j = 0; j < oldImg.getHeight(); j++){
//get the rgb color of the old image and store it the new
Color c = new Color(oldImg.getRGB(i, j));
int r = c.getRed();
int g = c.getGreen();
int b = c.getBlue();
int col = (r<<16) | (g<<8) | b;
newImg.setRGB(i, j, col);
}
}
//write image
f = new File("newImg.jpg");
ImageIO.write(newImg, "jpg", f);
}catch(IOException e){
System.out.println("Error: " + e);
}
}//main() ends here
}//class ends here
And I would like to basically slow the process down and display it happening. But I'm not sure how to do that. Would I need to use to accomplish this ? I'm somewhat new to threading but I think I would need multiple threads to handle the painting of both pictures.
First of all, I would like to mention you are working in a very inefficient way. You are creating a Color, decomposing a pixel in its channels, and moving to the new image by a bit-shift. It is easier if you work directly with the integer the whole time (and more efficient).
I will assume the image "/path/to/image/shrek4life.jpg" has ARGB color space. I recommend ensure this, because if the old image does not have this color space you should make a conversion.
When you create the new image, you create it as ARGB color space, so each channel is expressed in a byte of the int, first byte for Alpha, second byte for red, third byte for green and the last one for blue.
I think you forgot the alpha channel when you manipulated the old image pixel to move it into the new image.
With this explanation in mind, I think you can change your code to increase the efficiency, like this:
for(int i = 0; i < oldImg.getWidth(); i++){
for(int j = 0; j < oldImg.getHeight(); j++){
int pixel = oldImg.getRGB(i,j);
newImg.setRGB(i, j, pixel );
//If you like to get more control over the pixels and print
//you can decompose the pixel using Color as you already do
//But when you understand fully the process I recommend erase it
Color c = new Color(pixel);
//Print the color or do whatever you like
}
}
About how to display the process of pixel relocation:
In process:
You can print the changed pixel as a number with its position in image (discouraged). System.out.println("pixel"+pixel+" X:"+i+" Y:"+j);
Use this tutorial in baeldung to print an image. I suggest draw a rectangle with the color of the image and wait for a key press (enter, for example) using Scanner. After the key was press, you can load the next pixel, an so on.
If a single rectangle with just one pixel has little information, I suggest add an array of rectangles to draw several pixels in a time. Even you can print an image, and see the process pixel by pixel, using Scanner to mark each step.
As #haraldK suggest, you can use Swing to display de relocation image. Through swing timer and invokes update()
Post process:
Save the image in a file. To improve the speed of process, I suggest save a few pixels (10 - 100).

Speed up looking through matrix

I have code
public static void program() throws Exception{
BufferedImage input = null;
long start = System.currentTimeMillis();
while((System.currentTimeMillis() - start)/1000 < 220){
for (int i = 1; i < 13; i++){
for (int j = 1; j < 7; j++){
input = robot.createScreenCapture(new Rectangle(3+i*40, 127+j*40, 40, 40));
if ((input.getRGB(6, 3) > -7000000) && (input.getRGB(6, 3)<-5000000)){
robot.mouseMove(10+i*40, 137+j*40);
robot.mousePress(InputEvent.BUTTON1_MASK);
robot.mouseRelease(InputEvent.BUTTON1_MASK);
}
}
}
}
}
On a webpage there's a matrix (12*6) and there will randomly spawn some images. Some are bad, some are good.
I'm looking for a better way to check for good images. At the moment, on good images on location (6,3) the RGB color is different from bad images.
I'm making screenshot from every box (40 * 40) and looking at pixel in location (6,3)
Don't know how to explain my code any better
EDIT:
Picture of the webpage. External links ok?
http://i.imgur.com/B5Ev1Y0.png
I'm not sure what exactly the bottleneck is in your code, but I have a hunch it might be the repeated calls to robot.createScreenCapture.
You could try calling robot.createScreenCapture on the entire matrix (i.e. a large rectangle that covers all the smaller rectangles you are interested in) outside your nested loops, and then look up the pixel values at the points you are interested in using offsets for the x and y coordinates for the sub rectangles you are inspecting.

Java: Checking if image moved

I want to look within a certain position in an image to see if the selected pixels have changed in color, how would I go about doing this? (Im trying to check for movement)
I was thinking I could do something like this:
public int[] rectanglePixels(BufferdImage img, Rectangle Range) {
int[] pixels = ((DataBufferByte) bufferedImage.getRaster().getDataBuffer()).getData();
int[] boxColors;
for(int y = 0; y < img.getHeight(); y++) {
for(int x = 0; x < img.getWidth; x++) {
boxColors = pixels[(x & Range.width) * Range.x + (y & Range.height) * Range.y * width]
}
}
return boxColors;
}
Maybe use that to extract the colors from the position? Not sure if im doing that right, but after that should I re-run this method, compare the two arrays for similarities? and if the number of similarities reach some threshold declare that the image has changed?
One approach to detect movement is the analysis of pixel color variation considering the entire image or a subimage in distinct times (n, n-1, n-2, ...). In this case you are considering a fixed camera. You might have two thresholds:
The threshold of color channel variation that defines that two pixels are distinct.
The threshold of distinct pixels between the images to consider there is movement. In other words: two images of the same scene at time n and n-1 have just 10 distinct pixels. It is a real movement or just noise?
Below an example showing how to counter the distict pixels in an image, given a color channel threshold.
for(int y=0; y<imageA.getHeight(); y++){
for(int x=0; x<imageA.getWidth(); x++){
redA = imageA.getIntComponent0(x, y);
greenA = imageA.getIntComponent1(x, y);
blueA = imageA.getIntComponent2(x, y);
redB = imageB.getIntComponent0(x, y);
greenB = imageB.getIntComponent1(x, y);
blueB = imageB.getIntComponent2(x, y);
if
(
Math.abs(redA-redB)> colorThreshold ||
Math.abs(greenA-greenB)> colorThreshold||
Math.abs(blueA-blueB)> colorThreshold
)
{
distinctPixels++;
}
}
}
However, there are Marvin plug-ins to do so. Check this source code example. It detects and display regions containing "movements", as shown in the image below.
There are more sophisticated approaches that determine/subtract background for this purpose or deal with camera movements. I guess you should start from the simplest scenario and then go to more complex ones.
You should use BufferedImage.getRGB(startX, startY, w, h, rgbArray, offset, scansize) unless you really want to play around with the loops and extra arrays.
Comparing two values through a threshold would serve as good indicator. Perhaps, you could calculate averages for each array to determine color and compare the two? If you do not want a threshold value just use .hashCode();

What is the fastest way to read a large tiff image in java?

I currently use the JAI library to read the tiff image but it is very very slow large tiff images (I need to work with satellite images of size around 1GB).
I need to read the height of each point from the tiff image and then color it accordingly.
I am reading the image by creating a PlanarImage and iterating through every pixel by using the image.getData().getPixel(x,y,arr) method.
Suggest me a better way of implementing the solution.
Edit:
I found the error.I was creating a new raster of the image for every pixel by calling the image.getData() method in the for loop.Creating a raster just once and then using its getPixel() function in the loop solved my problem.
From the JavaDoc of PlanarImage.getData():
The returned Raster is semantically a copy.
This means that for every pixel of your image, you are creating a copy of the entire image in memory... This cannot give good performance.
Using getTile(x, y) or getTiles() should be faster.
Try:
PlanarImage image;
final int tilesX = image.getNumXTiles();
final int tilesY = image.getNumYTiles();
int[] arr = null;
for (int ty = image.getMinTileY(); ty < tilesY; ty++) {
for (int tx = startX; tx < image.getMinTileX(); tx++) {
Raster tile = image.getTile(tx, ty);
final int w = tile.getWidth();
final int h = tile.getHeight();
for (int y = tile.getMinY(); y < h; y++) {
for (int x = tile.getMinX(); x < w; x++) {
arr = tile.getPixel(x, y, arr);
// do stuff with arr
}
}
}
}
A 1 GB compressed image is likely to be about 20 GB+ when loaded into memory. The only way to handle this in Java is to load it with a very large heap space.
You are dealing with very large images and the simplest way to make this faster is to use a faster PC. I suggest an over clocked i7 3960X which you can get for a reasonable price http://www.cpubenchmark.net/high_end_cpus.html

Get Pixel Color around an image

I have an image, and I figured out how to use robot and getPixelColor() to grab the color of a certain pixel. The image is a character that I'm controlling, and I want robot to scan around the image constantly, and tell me if the pixels around it equal a certain color. Is this at all possible? Thanks!
Myself, I'd use the Robot to extract the image that's just a little larger than the "character", and then analyze the BufferedImage obtained. The details of course will depend on the details of your program. Probably the quickest would be to get the BufferedImage's Raster, then get thats dataBuffer, then get thats data, and analyze the array returned.
For example,
// screenRect is a Rectangle the contains your "character"
// + however many images around your character that you desire
BufferedImage img = robot.createScreenCapture(screenRect);
int[] imgData = ((DataBufferInt)img.getRaster().getDataBuffer()).getData();
// now that you've got the image ints, you can analyze them as you wish.
// All I've done below is get rid of the alpha value and display the ints.
for (int i = 0; i < screenRect.height; i++) {
for (int j = 0; j < screenRect.width; j++) {
int index = i * screenRect.width + j;
int imgValue = imgData[index] & 0xffffff;
System.out.printf("%06x ", imgValue );
}
System.out.println();
}

Categories

Resources