trying to record video from camera by using javaCV,
// recoder settings:
private int imageWidth = 320;
private int imageHeight = 240;
private int frameRate = 30;
recorder = new FFmpegFrameRecorder(ffmpeg_link, imageWidth, imageHeight, 1);
recorder.setFormat("mp4");
recorder.setFrameRate(frameRate);
// frame settings:
IplImage yuvIplimage = null;
yuvIplimage = IplImage.create(320, 320, IPL_DEPTH_16U, 1); //32 not supported
//image reader:
private ImageReader mImageReader;
mImageReader = ImageReader.newInstance(320, 320, ImageFormat.YUV_420_888, 10);
mImageReader.setOnImageAvailableListener(
mOnImageAvailableListener, mBackgroundHandler);
private final ImageReader.OnImageAvailableListener mOnImageAvailableListener
= new ImageReader.OnImageAvailableListener() {
#Override
public void onImageAvailable(ImageReader reader) {
Image image = reader.acquireNextImage();// acquireLatestImage(); - also tried
if (image == null)
return;
final ByteBuffer buffer = image.getPlanes()[0].getBuffer();
byte[] bytes = new byte[buffer.remaining()];
buffer.get(bytes, 0, bytes.length);
if (yuvIplimage != null ) {
// OPTION 1
yuvIplimage.getByteBuffer().put(convertYUV420ToNV21(image));
// OPTION 2
//yuvIplimage.getByteBuffer().put(decodeYUV420SP(bytes,320,320));
try {
if (started) {
recorder.record(yuvIplimage);
}
} catch (Exception e) {
e.printStackTrace();
}
}
image.close();
}
};
Option 1 is to decode image to NV21 by using code:
private byte[] convertYUV420ToNV21(Image imgYUV420) {
byte[] rez;
ByteBuffer buffer0 = imgYUV420.getPlanes()[0].getBuffer();
ByteBuffer buffer2 = imgYUV420.getPlanes()[2].getBuffer();
int buffer0_size = buffer0.remaining();
int buffer2_size = buffer2.remaining();
rez = new byte[buffer0_size + buffer2_size];
buffer0.get(rez, 0, buffer0_size);
buffer2.get(rez, buffer0_size, buffer2_size);
return rez;
}
option 2 is to convert to rgb as if i understand corect:
public byte[] decodeYUV420SP( byte[] yuv420sp, int width, int height) {
final int frameSize = width * height;
byte rgb[]=new byte[width*height];
for (int j = 0, yp = 0; j < height; j++) {
int uvp = frameSize + (j >> 1) * width, u = 0, v = 0;
for (int i = 0; i < width; i++, yp++) {
int y = (0xff & ((int) yuv420sp[yp])) - 16;
if (y < 0) y = 0;
if ((i & 1) == 0) {
v = (0xff & yuv420sp[uvp++]) - 128;
u = (0xff & yuv420sp[uvp++]) - 128;
}
int y1192 = 1192 * y;
int r = (y1192 + 1634 * v);
int g = (y1192 - 833 * v - 400 * u);
int b = (y1192 + 2066 * u);
if (r < 0) r = 0; else if (r > 262143) r = 262143;
if (g < 0) g = 0; else if (g > 262143) g = 262143;
if (b < 0) b = 0; else if (b > 262143) b = 262143;
rgb[yp] = (byte) (0xff000000 | ((r << 6) & 0xff0000)
| ((g >> 2) & 0xff00) | ((b >> 10) & 0xff));
}
}
return rgb; }
it looks incorect also.
wisch is the correct way to conert camera2 image to IplImage?
and is it possible to do it on fly?
If the recorder requires NV21, then converting the image to that instead of RGB is likely the fastest option.
But why don't you just use android.media.MediaRecorder? It's much more efficient and can use the hardware encoders.
But if you need to stick with ffmpeg, your first option is incorrect for many devices. In addition, make sure you remove that buffer.get call earlier - it'll make the rest of the reads from plane 0 not work right, which may be your current problem. Once you read plane 0 once, .remaining() will return 0.
The YUV image has 3 planes, and unless you've checked that the underlying format is actually NV21, you shouldn't blindly assume that., or assume that the row stride is equal to width.
To be safe, you need to look at both row and pixel strides when copying the three planes into the semiplanar byte[].
If you're willing to move your processing over to C++ you could do the conversion like
cv::cvtColor((cv::_InputArray)mNV, (cv::_OutputArray)rgba, CV_YUV2RGBA_NV12, 0);
This will store the image as an RGBA mat and you can save it as normal from there.
Related
In onPreviewFrame using android.hardware.Camera, if I use the default NV21 format, I can use YuvImage to compress to jpeg format, which works great. If I try to change the format using setPreviewFormat(ImageFormat.YV12), then it does not work anymore as YuvImage does not support YV12 format. I've found only one solution somewhere to convert Bitmap to YV12, but I want to do the opposite and get a jpeg out of these bytes. Is there a library to do this?
If YUV420 to JPEG is what you're looking for then,
#Override
public void onPreviewFrame(byte[] data, Camera camera) {
// TODO Auto-generated method stub
ByteArrayOutputStream out= new ByteArrayOutputStream();
decodeYUV420(pixels, data, previewSize.width, previewSize.height);
mBitmap = Bitmap.createBitmap(pixels, previewSize.width, previewSize.height,Config.ARGB_8888);
mBitmap.compress(CompressFormat.JPEG , 25, out);
.......
where, the decodeYUV420 method goes as follows:
public void decodeYUV420(int[] rgb, byte[] yuv420, int width, int height) {
final int frameSize = width * height;
for (int j = 0, yp = 0; j < height; j++) {
int uvp = frameSize + (j >> 1) * width, u = 0, v = 0;
for (int i = 0; i < width; i++, yp++) {
int y = (0xff & ((int) yuv420[yp])) - 16;
if (y < 0) y = 0;
if ((i & 1) == 0) {
v = (0xff & yuv420[uvp++]) - 128;
u = (0xff & yuv420[uvp++]) - 128;
}
int y1192 = 1192 * y;
int r = (y1192 + 1634 * v);
int g = (y1192 - 833 * v - 400 * u);
int b = (y1192 + 2066 * u);
if (r < 0) r = 0; else if (r > 262143) r = 262143;
if (g < 0) g = 0; else if (g > 262143) g = 262143;
if (b < 0) b = 0; else if (b > 262143) b = 262143;
rgb[yp] = 0xff000000 | ((r << 6) & 0xff0000) | ((g >> 2) & 0xff00) | ((b >> 10) & 0xff);
}
}
I've created a similar project to this on github check here
and the code implementation: here
and yes it works!
I am building a scanner app, and trying to determine the "preview quality" from the preview callback of the camera. I want to customize the camera's AUTO_FLASH_MODE where it will be turned on if the environment is too dark.
How can I detect if there is a high average of dark pixels? This means (in preview) I am getting darkness and therefore need to turn on the camera's flash light.
Either find out how to access pixel values of your image and calculate the average intensity yourself or use any image processing library to do so.
Dark pixels have low values, bright pixels have high values.
You want to calculate the average of all red, green and blue values divided by three times your pixel count.
Define a threshold for when to turn on the flash, but keep in mind that you have to get a new exposure time then.
Prefer flash over exposure time increase as long exposure times yield higher image noise.
I tried this approach but i think it is taking unnecessary time of processing the bitmap and then get an average screen color,
#Override
public void onPreviewFrame(byte[] data, Camera camera) {
Size cameraResolution = resolution;
PreviewCallback callback = this.callback;
if (cameraResolution != null && callback != null)
{
int format = camera.getParameters().getPreviewFormat();
SourceData source = new SourceData(data, cameraResolution.width, cameraResolution.height, format, getCameraRotation());
callback.onPreview(source);
final int[] rgb = decodeYUV420SP(data, cameraResolution.width, cameraResolution.height);
//Bitmap bmp = decodeBitmap(source.getData());
Bitmap bmp = Bitmap.createBitmap(rgb, cameraResolution.width, cameraResolution.height, Bitmap.Config.ARGB_8888);
if (bmp != null)
{
//bmp = decodeBitmap(source.getData());
ByteArrayOutputStream bytes = new ByteArrayOutputStream();
// bmp.compress(Bitmap.CompressFormat.JPEG, 70, bytes);
Bitmap resizebitmap = Bitmap.createBitmap(bmp,
bmp.getWidth() / 2, bmp.getHeight() / 2, 60, 60);
int color = getAverageColor(resizebitmap);
Log.i("Color Int", color + "");
// int color = resizebitmap.getPixel(resizebitmap.getWidth()/2,resizebitmap.getHeight()/2);
String strColor = String.format("#%06X", 0xFFFFFF & color);
//String colorname = sColorNameMap.get(strColor);
Log.d("strColor", strColor);
Log.i("strColor", color + "");
if(!mIsOn)
{
if (color == -16777216 || color < -16777216)//minimum color code (full dark)
{
mIsOn = true;
setTorch(true);
Log.d("Yahooooo", "" + color);
}
}
Log.i("Pixel Value",
"Top Left pixel: " + Integer.toHexString(color));
}
}
else
{
Log.d(TAG, "Got preview callback, but no handler or resolution available");
}
}
}
private int[] decodeYUV420SP(byte[] yuv420sp, int width, int height)
{
final int frameSize = width * height;
int rgb[]=new int[width*height];
for (int j = 0, yp = 0; j < height; j++) {
int uvp = frameSize + (j >> 1) * width, u = 0, v = 0;
for (int i = 0; i < width; i++, yp++) {
int y = (0xff & ((int) yuv420sp[yp])) - 16;
if (y < 0) y = 0;
if ((i & 1) == 0) {
v = (0xff & yuv420sp[uvp++]) - 128;
u = (0xff & yuv420sp[uvp++]) - 128;
}
int y1192 = 1192 * y;
int r = (y1192 + 1634 * v);
int g = (y1192 - 833 * v - 400 * u);
int b = (y1192 + 2066 * u);
if (r < 0) r = 0; else if (r > 262143) r = 262143;
if (g < 0) g = 0; else if (g > 262143) g = 262143;
if (b < 0) b = 0; else if (b > 262143) b = 262143;
rgb[yp] = 0xff000000 | ((r << 6) & 0xff0000) | ((g >> 2) &
0xff00) | ((b >> 10) & 0xff);
}
}
return rgb;
}
private int getAverageColor(Bitmap bitmap)
{
int redBucket = 0;
int greenBucket = 0;
int blueBucket = 0;
int pixelCount = 0;
for (int y = 0; y < bitmap.getHeight(); y++) {
for (int x = 0; x < bitmap.getWidth(); x++) {
int c = bitmap.getPixel(x, y);
pixelCount++;
redBucket += Color.red(c);
greenBucket += Color.green(c);
blueBucket += Color.blue(c);
// does alpha matter?
}
}
int averageColor = Color.rgb(redBucket / pixelCount, greenBucket
/ pixelCount, blueBucket / pixelCount);
return averageColor;
}
I get the images from device camera:
public synchronized void onPreviewFrame(byte[] data, Camera camera) {
//Get image from data
}
Decode the images to RGB:
int[] rgb = decodeYUV420SPtoRGB(data, width, height);
RGB to Bitmap:
Bitmap bitmap = rgbToBitmap(rgb,width, height);
Print images in ImageView to make sure that conversion is fine:
iv.setImageBitmap(bitmap);
Detect num faces:
FaceDetector faceDet = new FaceDetector(bitmap.getWidth(), bitmap.getHeight(), 1);
Face[] faceList = new Face[1];
int faces = faceDet.findFaces(bitmap, faceList);
Log.e("NumFaces: ",faces + "");
Decode functions:
private int[] decodeYUV420SPtoRGB(byte[] yuv420sp, int width, int height) {
if (yuv420sp == null) throw new NullPointerException();
final int frameSize = width * height;
int[] rgb = new int[frameSize];
for (int j = 0, yp = 0; j < height; j++) {
int uvp = frameSize + (j >> 1) * width, u = 0, v = 0;
for (int i = 0; i < width; i++, yp++) {
int y = (0xff & (yuv420sp[yp])) - 16;
if (y < 0) y = 0;
if ((i & 1) == 0) {
v = (0xff & yuv420sp[uvp++]) - 128;
u = (0xff & yuv420sp[uvp++]) - 128;
}
int y1192 = 1192 * y;
int r = (y1192 + 1634 * v);
int g = (y1192 - 833 * v - 400 * u);
int b = (y1192 + 2066 * u);
if (r < 0) r = 0;
else if (r > 262143) r = 262143;
if (g < 0) g = 0;
else if (g > 262143) g = 262143;
if (b < 0) b = 0;
else if (b > 262143) b = 262143;
rgb[yp] = 0xff000000 | ((r << 6) & 0xff0000) | ((g >> 2) & 0xff00) | ((b >> 10) & 0xff);
}
}
return rgb;
}
public static Bitmap rgbToBitmap(int[] rgb, int width, int height) {
if (rgb == null) throw new NullPointerException();
Bitmap bitmap = Bitmap.createBitmap(width, height, Bitmap.Config.RGB_565);
bitmap.setPixels(rgb, 0, width, 0, 0, width, height);
return bitmap;
}
I get always 0 in Log.e("NumFaces: ",faces + ""); While I'm seeing in the imageView faces, real faces and faces in photos. I tried it with physical devices, 2.3 (bad camera) and 4.2 (good camera HD)
Edit: in 4.2 device I am getting 800x480 frames in onPreviewFrame.
Any idea what happens?
I have solved the problem by holding the device in my hands horizontally instead of vertically.
I am using JavaCV in Android.
In my code, I have created a ImageComparator(class of OpenCV CookBook
http://code.google.com/p/javacv/source/browse/OpenCV2_Cookbook/src/opencv2_cookbook/chapter04/ImageComparator.scala?repo=examples
http://code.google.com/p/javacv/wiki/OpenCV2_Cookbook_Examples_Chapter_4) Object and use that object to compare images. If I use file from SD card the comparator is working.
File referenceImageFile = new File(absPath1); // Read an image.
IplImage reference = Util.loadOrExit(referenceImageFile,CV_LOAD_IMAGE_COLOR);
comparator = new ImageComparator(reference);
comparator = new ImageComparator(reference);
But from Camera Preview, when I am creating IplImage it is not working. I am getting the following Exception during comparison "score" calculation.
score = referenceComparator.compare(grayImage) / imageSize;
java.lang.RuntimeException: /home/saudet/android/OpenCV-2.4.2/modules/core/src/convert.cpp:1196: error: (-215) i < src.channels() in function void cvSplit(const void*, void*, void*, void*, void*)
For CameraPreview I am using the code from FacePreview to create IplImage.But it create Image in grayScale.
int f = SUBSAMPLING_FACTOR;
if (grayImage == null || grayImage.width() != width / f
|| grayImage.height() != height / f) {
grayImage = IplImage.create(width / f, height / f, IPL_DEPTH_8U, 1);
}
int imageWidth = grayImage.width();
int imageHeight = grayImage.height();
int dataStride = f * width;
int imageStride = grayImage.widthStep();
ByteBuffer imageBuffer = grayImage.getByteBuffer();
for (int y = 0; y < imageHeight; y++) {
int dataLine = y * dataStride;
int imageLine = y * imageStride;
for (int x = 0; x < imageWidth; x++) {
imageBuffer.put(imageLine + x, data[dataLine + f * x]);
}
}
How to create a Color IplImage from Camera to use with ImageComparator?
The below code seems to be working fine.
public void onPreviewFrame(final byte[] data, final Camera camera) {
try {
Camera.Size size = camera.getParameters().getPreviewSize();
processImage(data, size.width, size.height);
camera.addCallbackBuffer(data);
} catch (RuntimeException e) {
// The camera has probably just been released, ignore.
Log.d("Exception", " " + e);
}
}
protected void processImage(byte[] data, int width, int height) {
score.clear();
// First, downsample our image
int f = SUBSAMPLING_FACTOR;
IplImage _4image = IplImage.create(width, height, IPL_DEPTH_8U, f);
int[] _temp = new int[width * height];
if (_4image != null) {
decodeYUV420SP(_temp, data, width, height);
_4image.getIntBuffer().put(_temp);
}
//bitmap = Bitmap.createBitmap(width, height, Bitmap.Config.ARGB_8888);
//bitmap.copyPixelsFromBuffer(_4image.getByteBuffer());
Log.d("CompareAndroid", "processImage");
int imageSize = _4image.width() * _4image.height();
Iterator<ImageComparator> iterator = reference_List.iterator();
// Compute histogram match and normalize by image size.
// 1 means perfect match.
while(iterator.hasNext()){
score.add(((ImageComparator) iterator.next()).compare(_4image) / imageSize);
}
Log.d("CompareImages", "Score Size "+score.size());
postInvalidate();
}
This code seems to be working fine.
private void decodeYUV420SP(int[] rgb, byte[] yuv420sp, int width,
int height) {
int frameSize = width * height;
for (int j = 0, yp = 0; j < height; j++) {
int uvp = frameSize + (j >> 1) * width, u = 0, v = 0;
for (int i = 0; i < width; i++, yp++) {
int y = (0xff & ((int) yuv420sp[yp])) - 16;
if (y < 0)
y = 0;
if ((i & 1) == 0) {
v = (0xff & yuv420sp[uvp++]) - 128;
u = (0xff & yuv420sp[uvp++]) - 128;
}
int y1192 = 1192 * y;
int r = (y1192 + 1634 * v);
int g = (y1192 - 833 * v - 400 * u);
int b = (y1192 + 2066 * u);
if (r < 0)
r = 0;
else if (r > 262143)
r = 262143;
if (g < 0)
g = 0;
else if (g > 262143)
g = 262143;
if (b < 0)
b = 0;
else if (b > 262143)
b = 262143;
rgb[yp] = 0xff000000 | ((b << 6) & 0xff0000) | ((g >> 2) & 0xff00) | ((r >> 10) & 0xff);
}
}
}
I haven't tested it, but something like this should work:
IplImage yuvimage = IplImage.create(width, height * 3 / 2, IPL_DEPTH_8U, 2);
IplImage rgbimage = IplImage.create(width, height, IPL_DEPTH_8U, 3);
cvCvtColor(yuvimage, rgbimage, CV_YUV2BGR_NV21);
How can i read the preview frames from the actual camera source code? I am trying to modify the actual source code of camera application in android for reading the preview frames. Can anyone help me with this.
You shouldn't have to change the actual source code for reading frames. Implementing the Camera.PreviewCallback interface should suffice. It returns raw data from the camera.
Before messing with the source of the camera app try the example here:
CameraPreview
Then implement the Camera.PreviewCallback.
The captured frame comes in YUV420SP so you have to convert it to rgb in order to convert it into a colored bitmap and show it on screen.
Like this :
#Override
public void onPreviewFrame(byte[] data, Camera camera) {
int imageWidth = camera.getParameters().getPreviewSize().width ;
int imageHeight =camera.getParameters().getPreviewSize().height ;
int RGBData[] = new int[imageWidth* imageHeight];
byte[] mYUVData = new byte[data.length];
System.arraycopy(data, 0, mYUVData, 0, data.length);
decodeYUV420SP(RGBData, mYUVData, imageWidth, imageHeight);
Bitmap bitmap = Bitmap.createBitmap(imageWidth, imageHeight, Bitmap.Config.ARGB_8888);
bitmap.setPixels(RGBData, 0, imageWidth, 0, 0, imageWidth, imageHeight);
}
static public void decodeYUV420SP(int[] rgb, byte[] yuv420sp, int width, int height) {
final int frameSize = width * height;
for (int j = 0, yp = 0; j < height; j++) {
int uvp = frameSize + (j >> 1) * width, u = 0, v = 0;
for (int i = 0; i < width; i++, yp++) {
int y = (0xff & ((int) yuv420sp[yp])) - 16;
if (y < 0) y = 0;
if ((i & 1) == 0) {
v = (0xff & yuv420sp[uvp++]) - 128;
u = (0xff & yuv420sp[uvp++]) - 128;
}
int y1192 = 1192 * y;
int r = (y1192 + 1634 * v);
int g = (y1192 - 833 * v - 400 * u);
int b = (y1192 + 2066 * u);
if (r < 0) r = 0; else if (r > 262143) r = 262143;
if (g < 0) g = 0; else if (g > 262143) g = 262143;
if (b < 0) b = 0; else if (b > 262143) b = 262143;
rgb[yp] = 0xff000000 | ((r << 6) & 0xff0000) | ((g >> 2) & 0xff00) | ((b >> 10) & 0xff);
}
}
}