I've written a small application for personal use in macOs using java (the only language I know) that reads a directory and compare the images found in it with every other image.
Edit: As user dbush pointed out, here is the code I should've posted:
for (int i = 0; i < files.size(); i++) {
file1 = files.get(i).getAbsolutePath();
for (int j = i + 1; j < files.size(); j++) {
file2 = files.get(j).getAbsolutePath();
System.out.println("Comparing " + files.get(i).getName() + sp + files.get(j).getName());
System.out.println(Arrays.equals(extractBytes(file1), extractBytes(file2)));
}
}
private static byte[] extractBytes(String imageName) throws IOException {
// open image
File imgPath = new File(imageName);
BufferedImage bufferedImage = getSample(imgPath);
// get DataBufferBytes from Raster
WritableRaster raster = bufferedImage.getRaster();
DataBufferByte data = (DataBufferByte) raster.getDataBuffer();
return (data.getData());
}
private static BufferedImage getSample(File f) {
BufferedImage img = null;
Rectangle sourceRegion = new Rectangle(0, 0, 10, 10); // The region you want to extract
try {
ImageInputStream stream1 = ImageIO.createImageInputStream(f); // File or input stream
Iterator<ImageReader> readers = ImageIO.getImageReaders(stream1);
if (readers.hasNext()) {
ImageReader reader = readers.next();
reader.setInput(stream1);
ImageReadParam param = reader.getDefaultReadParam();
param.setSourceRegion(sourceRegion); // Set region
img = reader.read(0, param); // Will read only the region specified
}
} catch (IOException e) {
System.out.println(e);
}
return img;
}
It involves reading two images in a byte array and comparing both. The problem is that it takes a long time to do this, even if I sample a small portion of each image.
Should I re-write the same application in C or swift for better performance, considering I'll be running it on macOs from the terminal? Or it will do minimal difference.
Thanks a lot in advance!
Related
I am receiving a MultipartFile Spring object from rest controller. I am trying to convert any inage file to JPG image but I just need the byte array to save it on mongoDb
I found this code to do that
public boolean convertImageToJPG(InputStream attachedFile) {
try {
BufferedImage inputImage = ImageIO.read(attachedFile);
ByteArrayOutputStream byteArrayOutputStream = new ByteArrayOutputStream();
boolean result = ImageIO.write(inputImage, "jpg", byteArrayOutputStream);
return result;
} catch (IOException e) {
System.err.println("Error " + e);
}
return false;
}
But result as a false with not error, so ImageIO.write is not working
Also I found this to do the same but using File object, I don't want to create the file on directory, I just need the byte array
public static boolean convertFormat(String inputImagePath,
String outputImagePath, String formatName) throws IOException {
FileInputStream inputStream = new FileInputStream(inputImagePath);
FileOutputStream outputStream = new FileOutputStream(outputImagePath);
// reads input image from file
BufferedImage inputImage = ImageIO.read(inputStream);
// writes to the output image in specified format
boolean result = ImageIO.write(inputImage, formatName, outputStream);
// needs to close the streams
outputStream.close();
inputStream.close();
return result;
}
Testing
public class TestImageConverter {
public static void main(String[] args) {
String inputImage = "D:/Photo/Pic1.jpg";
String oututImage = "D:/Photo/Pic1.png";
String formatName = "PNG";
try {
boolean result = ImageConverter.convertFormat(inputImage,
oututImage, formatName);
if (result) {
System.out.println("Image converted successfully.");
} else {
System.out.println("Could not convert image.");
}
} catch (IOException ex) {
System.out.println("Error during converting image.");
ex.printStackTrace();
}
}
}
How can I solve my problem?
UPDATED SOLUTION (alternative with no need for Raster and ColorModel)
It had indeed bothered me that my older solution (see below) still required Rasters and ColorModels. I got challenged on my solution, so I spent some more time looking for alternatives. So the best thing I could come up with now is the following:
try {
final FileInputStream fileInputStream = new FileInputStream("dice.png");
final BufferedImage image = ImageIO.read(fileInputStream);
fileInputStream.close(); // ImageIO.read does not close the input stream
final BufferedImage convertedImage = new BufferedImage(image.getWidth(), image.getHeight(), BufferedImage.TYPE_INT_RGB);
convertedImage.createGraphics().drawImage(image, 0, 0, Color.WHITE, null);
final FileOutputStream fileOutputStream = new FileOutputStream("dice-test.jpg");
final boolean canWrite = ImageIO.write(convertedImage, "jpg", fileOutputStream);
fileOutputStream.close(); // ImageIO.write does not close the output stream
if (!canWrite) {
throw new IllegalStateException("Failed to write image.");
}
} catch (IOException e) {
e.printStackTrace();
}
I ended up with a copy of the BufferedImage as I did before. It does more or less the same thing, but you can actually reuse the ColorModel and Raster more easily.
drawImage() seems to take care of most of what I did before manually. And since it is standard java library code all the way, it seems indeed to be a better way.
Note that you end up with an Image of type BufferedImage.TYPE_INT_RGB. While it seems to work for the types jpg, png, and gif, I am not sure what will happen to other file formats or files with a different storage ColorModel - information might be lost (e.g. 4 color-channels to 3). For the mentioned types we don't need an alpha channel, even if we convert from gif or jpg to png (it will be Color.WHITE).
OLD SOLUTION
I was not happy with my first design and also it did not quite work the way it should have.
Therefore, I have created one from scratch. I ended up with a little converter for sRGB files. You can convert from png to jpg and vice versa (Edit: Added gif support also). If you want to handle other types feel free to extend this further. You can more or less add it the same way. It might work for other file types as well, but I have not tested them yet. Luckily, it seems that sRGB is quite common though.
Tbh. I have no idea how many combinations and variants (color palettes, precision, quality, b/w, etc.) you can produce or which common properties they share.
Maybe this is good enough for you. Maybe not. At least it was a nice exercise for me.
This solution is by no means perfect. The results looked okay-ish. The file-type conversion worked and the file-size is also smaller than the png.
try {
final String fileName = "dice.png";
final BufferedImage inputImage = ImageIO.read(new FileInputStream(fileName));
final boolean isSRGB = inputImage.getColorModel().getColorSpace().isCS_sRGB();
final String outputFormat = "gif";
if (!isSRGB) {
throw new IllegalArgumentException("Please provide an image that supports sRGB.");
}
final WritableRaster raster = createRaster(inputImage);
final ColorModel colorModel = createColorModel(inputImage);
final BufferedImage outputImage = new BufferedImage(colorModel, raster, colorModel.isAlphaPremultiplied(), null);
final String outputFileName = fileName + "-converted." + outputFormat;
final boolean writeResult = ImageIO.write(outputImage, outputFormat, new FileOutputStream(outputFileName));
if (!writeResult) {
throw new IllegalStateException("Could not convert file: " + fileName + " to format: " + outputFormat);
}
System.out.println(">> Created file: " + outputFileName);
} catch (Exception e) {
e.printStackTrace();
}
#NotNull
public static ColorModel createColorModel(#NotNull BufferedImage bufferedImage) {
Objects.requireNonNull(bufferedImage);
final int type = bufferedImage.getType();
boolean isAlphaPremultiplied = false;
int transparency = Transparency.OPAQUE;
if (type == BufferedImage.TYPE_3BYTE_BGR) {
isAlphaPremultiplied = true;
}
return new ComponentColorModel(
ColorModel.getRGBdefault().getColorSpace(),
false, isAlphaPremultiplied, transparency,
bufferedImage.getData().getDataBuffer().getDataType()
);
}
#NotNull
public static WritableRaster createRaster(#NotNull BufferedImage bufferedImage) {
Objects.requireNonNull(bufferedImage);
final int type = bufferedImage.getType();
final int width = bufferedImage.getWidth();
final int height = bufferedImage.getHeight();
final int pixelStride = 3;
int[] offset = new int[]{0, 1, 2};
DataBufferByte dataBufferByte;
if (type == BufferedImage.TYPE_4BYTE_ABGR || type == BufferedImage.TYPE_BYTE_INDEXED) {
int dataIndex = 0;
final byte[] data = new byte[height * width * pixelStride];
final int bitmask = 0xff;
for (int y = 0; y < height; y++) {
for (int x = 0; x < width; x++) {
final int rgb = bufferedImage.getRGB(x, y);
final int blue = bitmask & rgb;
final int green = bitmask & (rgb >> 8);
final int red = bitmask & (rgb >> 16);
if (rgb == 0) {
data[dataIndex++] = (byte) bitmask;
data[dataIndex++] = (byte) bitmask;
data[dataIndex++] = (byte) bitmask;
} else {
data[dataIndex++] = (byte) red;
data[dataIndex++] = (byte) green;
data[dataIndex++] = (byte) blue;
}
}
}
dataBufferByte = new DataBufferByte(data, data.length);
} else if (type == BufferedImage.TYPE_3BYTE_BGR) {
dataBufferByte = (DataBufferByte) bufferedImage.getRaster().getDataBuffer();
offset = new int[]{2, 1, 0};
} else {
throw new IllegalArgumentException("Cannot create raster for unsupported image type.");
}
return Raster.createInterleavedRaster(
dataBufferByte, width, height,
pixelStride * width, pixelStride,
offset,
null
);
}
EDIT: Added support for gif.
i only got the image data with NO header informations but i know several things like:
16 bit grayscale data
1200x1200 (although its 1200x900 but its likely to have a "bar" at the buttom)
the data are 2880000 bytes in size which fits 1200x1200 x 2bytes ->short
here is the raw image data
for visualizing i use this:
public static void saveImage(short[] pix, int width, int height, File outputfile) {
ColorSpace cs = ColorSpace.getInstance(ColorSpace.CS_GRAY);
int[] nBits = {16};
ComponentColorModel cm = new ComponentColorModel(cs, nBits,false, false, Transparency.OPAQUE, DataBuffer.TYPE_USHORT);
SampleModel sm = cm.createCompatibleSampleModel(width, height);
DataBufferShort db = new DataBufferShort(pix, width*height);
WritableRaster raster = Raster.createWritableRaster(sm, db, null);
BufferedImage bf = new BufferedImage(cm, raster, false, null);
if(outputfile!=null)
try {
if(!ImageIO.write(bf, "png", outputfile)) System.out.println("No writer found.");
System.out.println("Saved: "+outputfile.getAbsolutePath());
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
else System.out.println("error");
}
The data are read like this (only experimental/bad code, its only for testing):
for(int tt=1; tt<20; tt++) {
pix = new short[1200*1200];
i = 0;
int z = 0;
int line = 0;
//loop
while(i<(1200*1200)) {
pix[i++] = buffer.getShort(z);
z += tt;
if(z>=(len-1)) {
line += 2;
z = line;
if(z>=(len-1)) {
System.out.println("break at "+z);
break;
}
System.out.println("test "+line);
}
}
System.out.println("img_"+imgcount+".png "+pix.length);
saveImage(pix, 1200, 1200, new File("img_"+imgcount+"_"+tt+".png"));
}
Where i can see something for tt=4,8,16 (images get multiplied) but i cant realy get the whole picture.image tt=8 image tt=16
Its like the solution is in front of me but i cant see it xD
Can someone help me with the algorithm/format the image is stored?
EDIT: Reading data consecutively with:
short[] pix = new short[1200*1200];
int i = 0;
while(i< (1200*1200) && buffer.remaining()>0) {
pix[i++] = buffer.getShort();
}
results in: noisy picture
EDIT 2:
Ok looks like its base64 encoded which makes sense due its stored in a xml file
I finaly solved it, its base64 encoded and little endian (thanks RealSkeptic for hinting to try little/big endian).
I am trying to compare an embedded image located in a MP3 file with the exact same image saved as a JPG. If the images are identical then I would like to perform some further processing, however, when I compare the 2 images (RGB comparison) I keep getting false.
I am sure that the images are identical because I extracted the image from the same MP3 to originally create the JPG using the following code.
Artwork aw = tag.getFirstArtwork();
ByteArrayInputStream bis = new ByteArrayInputStream(aw.getBinaryData());
BufferedImage imgA = ImageIO.read(bis);
File outputfile = new File("expectedImage.jpg");
ImageIO.write(imgA, "jpg", outputfile);
After I ran that to get the image I just commented out that section, now I have the following code in place to compare the MP3 embedded image with the JPG
Extract the MP3 image and call the comparison method
try {
Artwork aw = tag.getFirstArtwork();
ByteArrayInputStream bis = new ByteArrayInputStream(aw.getBinaryData());
BufferedImage imgA = ImageIO.read(bis);
File expectedImageFile = new File("expectedImage.jpg");
BufferedImage imgB = ImageIO.read(expectedImageFile);
if(compareImages(imgA, imgB)) {
System.out.println("The Images Match.");
}else {
System.out.println("The Images Do Not Match.");
}
}
catch (IOException e) {
e.printStackTrace();
}
Compare The Images
The method fails when comparing the pixels for equality on the first pass through the loop.
public static boolean compareImages(BufferedImage imgA, BufferedImage imgB) {
// The images must be the same size.
if (imgA.getWidth() != imgB.getWidth() || imgA.getHeight() != imgB.getHeight()) {
return false;
}
int width = imgA.getWidth();
int height = imgA.getHeight();
// Loop over every pixel.
for (int y = 0; y < height; y++) {
for (int x = 0; x < width; x++) {
// Compare the pixels for equality.
if (imgA.getRGB(x, y) != imgB.getRGB(x, y)) {
return false;
}
}
}
return true;
}
I try you code and read the #Sami Kuhmonen comment and I understand.
When you use ImageIO.write(imgA, "jpg", outputfile) you pass by an other encoder and you can loss data.
You need change this by the standard technique.
Example [UPDATED]
public static void main(String[] args) {
try {
//Write the picture to BAOS
ByteArrayOutputStream byteBuffer = new ByteArrayOutputStream();
byteBuffer.write(aw.getBinaryData(), 0, 1);
//New file target
File outputfile = new File("expectedImage.jpg");
OutputStream outputStream = new FileOutputStream(outputfile);
byteBuffer.writeTo(outputStream);
outputStream.close();
File expectedImageFile = new File("expectedImage.jpg");
ByteArrayInputStream bis = new ByteArrayInputStream(aw.getBinaryData());
BufferedImage imgA = ImageIO.read(bis);
FileInputStream fis = new FileInputStream(expectedImageFile);
BufferedImage imgB = ImageIO.read(fis);
if(compareImages(imgA, imgB)) {
System.out.println("The Images Match.");
}else {
System.out.println("The Images Do Not Match.");
}
} catch (IOException e) {
e.printStackTrace();
}
}
public static boolean compareImages(BufferedImage imgA, BufferedImage imgB) {
// The images must be the same size.
if (imgA.getWidth() != imgB.getWidth() || imgA.getHeight() != imgB.getHeight()) {
return false;
}
int width = imgA.getWidth();
int height = imgA.getHeight();
// Loop over every pixel.
for (int y = 0; y < height; y++) {
for (int x = 0; x < width; x++) {
// Compare the pixels for equality.
if (imgA.getRGB(x, y) != imgB.getRGB(x, y)) {
return false;
}
}
}
return true;
}
I've been a little busy lately, but I finally got a chance to look back into this today and based on what was said above I figured that the main issue was that the image was being compressed in one case and in the other it wasn't. I am sure there is a better solution but a quick dirty fix was to just save a temporary JPEG version of the image that I would like to check for, then do the comparison using 2 buffered images, one that reads in the JPEG file I have saved to the directory and one that reads in the temp JPEG I just created and when the test is complete just use the File.delete() to remove the temp file.
Extract the MP3 image and call the comparison method
The Compare method is the same as originally stated
Artwork aw = tag.getFirstArtwork();
ByteArrayInputStream bis = new ByteArrayInputStream(aw.getBinaryData());
BufferedImage tempImg = ImageIO.read(bis);
File tempFile = new File("temp.jpg");
ImageIO.write(tempImg, "jpg", tempFile);
BufferedImage imgA = ImageIO.read(tempFile);
File expectedImageFile = new File("imgToCheckAgainst.jpg");
BufferedImage imgB = ImageIO.read(expectedImageFile);
if(compareImages(imgA, imgB)) {
System.out.println("The Images Match");
}else {
System.out.println("The images do not match.");
}
tempFile.delete();
import javax.imageio.ImageIO;
import org.bytedeco.javacv.FFmpegFrameGrabber;
public class FrameData
{
int count = 0;
int picWidth;
int picHeight;
BufferedImage img = null;
//GET FRAME COUNT
public int gf_count(int numofFrames, BufferedImage[] frameArray, String fileLocationsent, String videoNamesent) throws IOException
{
String fileLocation = fileLocationsent;
String videoName = videoNamesent;
int frameNums = numofFrames;
int totFrames = 0;
FFmpegFrameGrabber grab = new FFmpegFrameGrabber(fileLocation + videoName);
try { grab.start(); }
catch (Exception e) { System.out.println("Unable to grab frames"); }
for(int i = 0 ; i < frameNums ; i++)
{
try
{
frameArray[i]= grab.grab().getBufferedImage();
totFrames = i;
File outputfile = new File(fileLocation + "GrayScaledImage" + i + ".jpg");
ImageIO.write(frameArray[i], "jpg", outputfile);
}
catch (Exception e) { /*e.printStackTrace();*/ }
}//END for
return totFrames;
}//END METHOD long getFrameCount()
Hope someone can explain this to me...
I am just learning java so here goes...
I wrote this code to count the number of frames in a .mov file and to test my buffered image array I generated files of the images. As the code is, it works as planned... The problem is immediately after the capturing, if I send the bufferedimages out as files, they all seem to be just the first image. see example below...
for(int i = 0 ; i < frameNums ; i++)
{
try
{
frameArray[i]= grab.grab().getBufferedImage();
totFrames = i;
File outputfile = new File(fileLocation + "GrayScaledImage" + i + ".jpg");
ImageIO.write(frameArray[i], "jpg", outputfile);
}
catch (Exception e) { /*e.printStackTrace();*/ }
}//END for
And now if I change that to...
for(int i = 0 ; i < frameNums ; i++)
{
try
{
frameArray[i]= grab.grab().getBufferedImage();
totFrames = i; catch (Exception e) { /*e.printStackTrace();*/ }}
for(int j = 0; j < frameNums; j++)
{
File outputfile = new File(fileLocation + "GrayScaledImage" + j + ".jpg");
ImageIO.write(frameArray[j], "jpg", outputfile);
}
I don't understand why I am getting the same image repeatedly.
If further information Is required, just lemme know, this is my first programming question online... Usually find what I am looking for that others have asked. Couldn't find this one.
Thanks for your time
Ken
The problem is that the grab().getBufferedImage() does its work in the same buffer every time. When you assign a reference to that buffer in your loop, you are assigning a reference to the same buffer numofFrames times. What you are writing then is not the first frame, but the last frame. In order to fix this you need to do a "deep copy" of the BufferedImage. See code below:
public class FrameData {
BufferedImage img;
Graphics2D g2;
// GET FRAME COUNT
public int gf_count(int numFrames, BufferedImage[] frameArray, String fileLocation, String videoName) throws Exception, IOException {
Java2DFrameConverter converter = new Java2DFrameConverter();
int totFrames = 0;
img = new BufferedImage(100, 50, BufferedImage.TYPE_INT_ARGB);
g2 = img.createGraphics();
FFmpegFrameGrabber grab = new FFmpegFrameGrabber(fileLocation + videoName);
grab.start();
for (int i = 0; i < numFrames; i++) {
frameArray[i] = deepCopy(converter.convert(grab.grab()));
totFrames = i;
}
for (int j = 0; j < totFrames; j++) {
File outputfile = new File(fileLocation + "TImage" + j + ".jpg");
ImageIO.write(frameArray[j], "jpg", outputfile);
}
return totFrames;
}// END METHOD long getFrameCount()
BufferedImage deepCopy(BufferedImage bi) {
ColorModel cm = bi.getColorModel();
boolean isAlphaPremultiplied = cm.isAlphaPremultiplied();
WritableRaster raster = bi.copyData(null);
return new BufferedImage(cm, raster, isAlphaPremultiplied, null);
}
// This does what the converter.convert seems to do, which
// is decode an image into the same place over and over.
// if you don't copy the result every time, then you end up
// with an array of references to the same last frame.
BufferedImage draw() {
g2.setColor(new Color(-1));
g2.fillRect(0, 0, 100, 50);
g2.setColor(new Color(0));
g2.drawLine(0, 0, (int)(Math.random()*100.0), (int)(Math.random()*50.0));
return img;
}
public static void main(String... args) throws Exception, IOException {
new FrameData().run();
}
private void run() throws Exception, IOException {
BufferedImage images[] = new BufferedImage[50];
gf_count(50, images, "C:/Users/karl/Videos/", "dance.mpg");
}
}
I have included a draw() method that shows by example how work is done in the same BufferedImage repeatedly, in case you want to replicate the problem.
There are certainly other ways to do a deep copy and there may be issues with the one shown. Reference: How do you clone a BufferedImage.
PS> I updated the code to use the 1.1 version of the bytedeco library.
So I have a very small piece of code, which takes a .gif image as input, and then it splits this .gif image into an array of BufferedImage. After that, it stores the images in the array on the disk. When I do this, the output images contain heavy white-pixel noise which isn't visible in the input .gif image.
Example of the input gif:
Example of malformed output image (the 3rd frame in the gif):
The code I am using to split the gif is as follows:
public static void main(String[] args) throws Exception {
splitGif(new File("C:\\test.gif"));
}
public static void splitGif(File file) throws IOException {
ImageReader reader = ImageIO.getImageReadersBySuffix("gif").next(); reader.setInput(ImageIO.createImageInputStream(new FileInputStream(file)), false);
for(int i = 0; i < reader.getNumImages(true); i++) {
BufferedImage image = reader.read(i);
ImageIO.write(image, "PNG", new File(i + ".png"));
}
}
Can anyone help me out?
So the problem was that when reading .gif files into java, then all pixels in a given frame that did not change color compared to their previous frame will be fully transparent. If you want to read a .gif and split it in an array of properly rendered BufferedImages, then you have to fill the transparent pixels of the current frame with the last non-transparent pixel of one of the previous frames.
Code:
public static void splitGif(File file) throws IOException {
ImageReader reader = ImageIO.getImageReadersBySuffix("gif").next();
reader.setInput(ImageIO.createImageInputStream(new FileInputStream(file)), false);
BufferedImage lastImage = reader.read(0);
ImageIO.write(lastImage, "PNG", new File(0 + ".png"));
for (int i = 1; i < reader.getNumImages(true); i++) {
BufferedImage image = makeImageForIndex(reader, i, lastImage);
ImageIO.write(image, "PNG", new File(i + ".png"));
}
}
private static BufferedImage makeImageForIndex(ImageReader reader, int index, BufferedImage lastImage) throws IOException {
BufferedImage image = reader.read(index);
BufferedImage newImage = new BufferedImage(image.getWidth(), image.getHeight(), BufferedImage.TYPE_4BYTE_ABGR);
if(lastImage != null) {
newImage.getGraphics().drawImage(lastImage, 0, 0, null);
}
newImage.getGraphics().drawImage(image, 0, 0, null);
return newImage;
}
You solved it yourself, but for good order: every next frame is the accumulation of all prior frames filling up the transparent pixels in the current frame.
public static void splitGif(File file) throws IOException {
ImageReader reader = ImageIO.getImageReadersBySuffix("gif").next();
reader.setInput(ImageIO.createImageInputStream(new FileInputStream(file)), false);
BufferedImage outImage = null;
Graphics2D g = null;
for (int i = 0; i < reader.getNumImages(true); i++) {
BufferedImage image = reader.read(i);
if (g == null) {
BufferedImage outImage = new BufferedImage(image.getWidth(), image.getHeight(),
BufferedImage.TYPE_4BYTE_ABGR);
g = (Graphics2D) outImage.getGraphics();
}
g.drawImage(lastImage, 0, 0, null);
ImageIO.write(outImage, "PNG", new File(i + ".png"));
}
if (g != null) {
g.dispose();
}
}
getGraphics==createGraphics should be balanced be a dispose as documented.