在Android中从视频图像中获取帧

Question

I've implemented a simple application which shows the camera picture on the screen. What I like to do now is grab a single frame and process it as bitmap. From what I could find out to this point it is not an easy thing to do.

I've tried using the onPreviewFrame method with which you get the current frame as a byte array and tried to decode it with the BitmapFactory class but it returns null. The format of the frame is a headerless YUV which could be translated to bitmap but it takes too long on a phone. Also I've read that the onPreviewFrame method has contraints on the runtime, if it takes too long the application could crash.

So what is the right way to do this?

Answer 1

Ok what we ended up doing is using the onPreviewFrame method and decoding the data in a seperate Thread using a method which can be found in the android help group.

decodeYUV(argb8888, data, camSize.width, camSize.height);
Bitmap bitmap = Bitmap.createBitmap(argb8888, camSize.width,
                    camSize.height, Config.ARGB_8888);

...

// decode Y, U, and V values on the YUV 420 buffer described as YCbCr_422_SP by Android 
// David Manpearl 081201 
public void decodeYUV(int[] out, byte[] fg, int width, int height)
        throws NullPointerException, IllegalArgumentException {
    int sz = width * height;
    if (out == null)
        throw new NullPointerException("buffer out is null");
    if (out.length < sz)
        throw new IllegalArgumentException("buffer out size " + out.length
                + " < minimum " + sz);
    if (fg == null)
        throw new NullPointerException("buffer 'fg' is null");
    if (fg.length < sz)
        throw new IllegalArgumentException("buffer fg size " + fg.length
                + " < minimum " + sz * 3 / 2);
    int i, j;
    int Y, Cr = 0, Cb = 0;
    for (j = 0; j < height; j++) {
        int pixPtr = j * width;
        final int jDiv2 = j >> 1;
        for (i = 0; i < width; i++) {
            Y = fg[pixPtr];
            if (Y < 0)
                Y += 255;
            if ((i & 0x1) != 1) {
                final int cOff = sz + jDiv2 * width + (i >> 1) * 2;
                Cb = fg[cOff];
                if (Cb < 0)
                    Cb += 127;
                else
                    Cb -= 128;
                Cr = fg[cOff + 1];
                if (Cr < 0)
                    Cr += 127;
                else
                    Cr -= 128;
            }
            int R = Y + Cr + (Cr >> 2) + (Cr >> 3) + (Cr >> 5);
            if (R < 0)
                R = 0;
            else if (R > 255)
                R = 255;
            int G = Y - (Cb >> 2) + (Cb >> 4) + (Cb >> 5) - (Cr >> 1)
                    + (Cr >> 3) + (Cr >> 4) + (Cr >> 5);
            if (G < 0)
                G = 0;
            else if (G > 255)
                G = 255;
            int B = Y + Cb + (Cb >> 1) + (Cb >> 2) + (Cb >> 6);
            if (B < 0)
                B = 0;
            else if (B > 255)
                B = 255;
            out[pixPtr++] = 0xff000000 + (B << 16) + (G << 8) + R;
        }
    }

}

Link: http://groups.google.com/group/android-developers/browse_thread/thread/c85e829ab209ceea/3f180a16a4872b58?lnk=gst&q=onpreviewframe#3f180a16a4872b58

Answer 2

In API 17+, you can do conversion to RGBA888 from NV21 with the 'ScriptIntrinsicYuvToRGB' RenderScript. This allows you to easily process preview frames without manually encoding/decoding frames:

@Override 
public void onPreviewFrame(byte[] data, Camera camera) { 
   Bitmap bitmap = Bitmap.createBitmap(r.width(), r.height(), Bitmap.Config.ARGB_8888);
    Allocation bmData = renderScriptNV21ToRGBA888(
        mContext,
        r.width(),
        r.height(),
        data);
    bmData.copyTo(bitmap);
}

public Allocation renderScriptNV21ToRGBA888(Context context, int width, int height, byte[] nv21) {
  RenderScript rs = RenderScript.create(context);
  ScriptIntrinsicYuvToRGB yuvToRgbIntrinsic = ScriptIntrinsicYuvToRGB.create(rs, Element.U8_4(rs));

  Type.Builder yuvType = new Type.Builder(rs, Element.U8(rs)).setX(nv21.length);
  Allocation in = Allocation.createTyped(rs, yuvType.create(), Allocation.USAGE_SCRIPT);

  Type.Builder rgbaType = new Type.Builder(rs, Element.RGBA_8888(rs)).setX(width).setY(height);
  Allocation out = Allocation.createTyped(rs, rgbaType.create(), Allocation.USAGE_SCRIPT);

  in.copyFrom(nv21);

  yuvToRgbIntrinsic.setInput(in);
  yuvToRgbIntrinsic.forEach(out);
  return out;
}

Answer 3

I actually tried the code given the previous answer found that the Colorvalues are not exact. I checked it by taking both the preview and the camera.takePicture which directly returns a JPEG array. And the colors were very different. After a little bit more searching I found another example to convert the PreviewImage from YCrCb to RGB:

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);
        }
    }
}

The color values given by this and the takePicture() exactly match. I thought I should post it here. This is where I got this code from. Hope this helps.

Answer 4

Tim's RenderScript solution is great. Two comments here though:

1. Create and reuse RenderScript rs , and Allocation in, out . Creating them every frame will hurt the performance.
2. RenderScript support library can help you back support to Android 2.3.

Answer 5

I don't see any of the answers better in performance than the built-in way to convert it. You can get the bitmap using this.

        Camera.Parameters params = camera.getParameters();
        Camera.Size previewsize = params.getPreviewSize();
        YuvImage yuv = new YuvImage(data, ImageFormat.NV21, previewsize.width, previewsize.height, null);
        ByteArrayOutputStream stream = new ByteArrayOutputStream();
        
        yuv.compressToJpeg(new Rect(0,0,previewsize.width, previewsize.height), 100, stream);

        byte[] buf = stream.toByteArray();
        Bitmap bitmap = BitmapFactory.decodeByteArray(buf, 0, buf.length);