I'm trying to find a method, way or kind of algorithm to downsize YUV images to an specific width and height values without converting it into RGB, just manipulating the byte[] of my YUV image.
I just found another topics about this as Resize (downsize) YUV420sp image
I see that the way to achieve this is removing pixels, but I can just do it always using a factor of 4 because chroma pixels are shared between 4 luma pixels.
After researching I just achieve the next method, which rescale a YUV image four times smaller than the original image, but want I want to achieve is freely convert from a Width x Height resolution to a smaller one that I want, not a factor of 4. It is possible to achieve this in some way? I don't have problems in use Renderscript or any kind of libraries.
/**
* Rescale a YUV image four times smaller than the original image for faster processing.
*
* @param data Byte array of the original YUV image
* @param imageWidth Width in px of the original YUV image
* @param imageHeight Height in px of the original YUV image
* @return Byte array containing the downscaled YUV image
*/
public static byte[] quadYuv420(byte[] data, int imageWidth, int imageHeight) {
Log.v(TAG, "[quadYuv420] receiving image with " + imageWidth + "x" + imageHeight);
long startingTime = System.currentTimeMillis();
byte[] yuv = new byte[imageWidth / 8 * imageHeight / 8 * 3 / 2];
// process Y component
int i = 0;
for (int y = 0; y < imageHeight; y += 8) {
for (int x = 0; x < imageWidth; x += 8) {
yuv[i] = data[y * imageWidth + x];
i++;
}
}
// process U and V color components
for (int y = 0; y < imageHeight / 2; y += 8) {
for (int x = 0; x < imageWidth; x += 16) {
if (i < yuv.length) {
yuv[i] = data[(imageWidth * imageHeight) + (y * imageWidth) + x];
i++;
yuv[i] = data[(imageWidth * imageHeight) + (y * imageWidth) + (x + 1)];
i++;
}
}
}
Log.v(TAG, "[quadYuv420] Rescaled YUV420 in " + (System.currentTimeMillis() - startingTime) + "ms");
return yuv;
}
