Question

I have a png file created with python PIL that contains an height map (positive values).

The format is : single channel (grey level) at 16bit, thus 16 bit per pixel.

I read the file on android with BitmapFactory.decodeStream(<...data input stream...>);

I correctly get the size of the image through getWidth() and getHeight().

However when I loop though the pixels calling the getPixel(i,j) I obtain negative values, something like: -16776192 -16250872 -16250872 -16250872 -16250872 -16249848 ….

Instead I expect a positive value between 0 and 65535.

I discover that the value -16250872 in binary is 1111111111111111111111111111111111111111000010000000100000001000

This shows that the information relies on the first 16 least significant bits.

I tryed with getPixel(i,j)&0xffff and I obtain a plausible values, however I’m not sure about the endianess: should I to flip the 2 extracted bytes?

Is there a way to do it in more elegant and portable way this conversion ?

NOTE: the file is not a color (RGBA) PNG but a grey level PNG image with a single 16 bit value for each pixel.

Answer 1

I found a solution by myself using the consideration made in this post:
Android: loading an alpha mask bitmap

Basically, If you load directly a 16 bit greylevel PNG from bitmap factory the pixel format will be not correct. You need to use a RGBA 32 bit color format setting the pixel format to ARGB_8888. Then you have to get all the pixels with getPixel(i,j) and mask the integer value with 0xffff. In this way you will obtain the expected 16bit values.

This is a part of the code that I’ve used:

BitmapFactory.Options opt = new BitmapFactory.Options();
opt.inPreferredConfig= Bitmap.Config.ARGB_8888;
Bitmap bmp=BitmapFactory.decodeStream(entity.getContent(),null,opt);
int W=bmp.getWidth();
int H=bmp.getHeight();
int px;
for (int i = 0; i < H; i++)
    for (int j = 0; j < W; j++)
    {
          px= bmp.getPixel(j, i)&0x0000ffff;//in px you will find a value between 0 and 65535
          ...
    }