MATLAB: Does IMREAD return more significant bits than are included in the PNG-file

bit depthbitdepthMATLAB

I have a PNG-file that has 16-bit color depth. The original image was 10-bit, and the file has been saved with the sBIT tag to indicate this.

However, when I use IMREAD to load this into MATLAB, the data has more then 16-bits. It appears the trailing bits, which I expected to be zero, repeat some of the significant bits.

Best Answer

In cases where the image has fewer significant bits than the bit depth of the PNG-file, the image encoder that produces the PNG-file must store the original significant bits as the most significant bits of the pixel data, but has the option of filling in the (undefined) least significant bits as it wishes, according to the PNG standard.

One common method of doing this is "left-bit replication", which you have observed. This has the effect of scaling the original data (approximately linearly) to fill the dynamic range allowed by the bit depth of the PNG-file. This accounts for the data repetition observed.

The image decoder has the option of ignoring the number of significant bits, but the original data can always be obtained by bit-shifting out the least significant bits.

You may use the BITSHIFT function in MATLAB to recover the original data. For instance, with the 10-bit image with 16-bit color depth:

data = imread('foo.png');
data = bitshift(data, 10 - 16);

Related Solutions

MATLAB: What is lsb planes of an image and what is its significance

"LSB" is Least Significant Bit.

Suppose you had 8 bits per pixel, 0 to 255. Then the Most Significant Bit (MSB) would be the bit that was either 0 or 128. A change in the MSB would make a difference of up to half of the available brightness. For example, 17 compared to 145 is a change in just the MSB. The Least Significant Bit (LSB) is the other end, the bit which can only make a difference of 1; for example 16 vs 17 is a change in the least significant bit

147 decimal = 10010001 binary
              |  |   |
              |  |   |--- contributes 1 out of the 147, least significant
              |  |------- contributes 16 out of the 147, intermediate significance
              |---------- contributes 128 out of the 147, most significant

When discussing LSB, the topic is often about steganography or watermarking. In those two very closely related topics, the significance of LSB is that if you need to change a bit of the data, then the LSB is the one that produces the least visual change, and so would be least likely to be noticed by the user.

Have look at

image(uint8(2.^(0:7)));colormap(gray(256))

the left bar has only the LSB set; the second bar has only the second bit set; the third bar has only the third bit set, and so on to the right that has most significant bit set. On the typical display settings that I use, I can clearly make out bar 6, 7, and 8, but I have a difficult time determining the boundary between bar 4 and bar 5, with 1 through 4 completely indistinguishable; even if I turn up to full brightness I have a hard time telling where the boundary between 4 and 5 is. So a change to bar 1, the least significant of them, would be visually invisible to me.

MATLAB: Convert an image with 16 bit depth to 32 or 62 bit in MATLAB

To convert a 16-bit monocromatic image (no colors, black and white) to 32-bit, the below procedure can be followed with gray scale image as reference. Here, Image.png is an RGB image.

Gray = imread('Image.png');
Gray = rgb2gray(Gray);
imwrite(Gray,'imagegrey.png');
grey_info = imfinfo('imagegrey.png');
Grey_BitDepth = grey_info.BitDepth
imwrite(Gray,'imagegrey16.png','BitDepth', 16);
grey16_info = imfinfo('imagegrey16.png');
Grey16_BitDepth = grey16_info.BitDepth

In the above code, maximum valid bit depth for a gray scale image is 16. Also, bit depths 24, 32 and 48 are supported for RGB images in MATLAB. A grayscale image is monochrome. Use this link for reference. https://www.mathworks.com/matlabcentral/answers/78417-can-i-convert-one-image-in-grayscale-to-monochrome

Bit depth of an image:

The image has a bit depth that is shown in the properties of original image under details tab in the file manager i.e. 32 as MATLAB supports only few bit depths that are specific to image file format. An image which has 1456*1936 unit16 value implies that the image has 16 bits/sample. Number of samples for a gray scale image is one and number of samples for rgb image are three.

Use the given link for reference.

https://www.mathworks.com/help/matlab/ref/imread.html

Best Answer

Related Solutions

MATLAB: What is lsb planes of an image and what is its significance

MATLAB: Convert an image with 16 bit depth to 32 or 62 bit in MATLAB

Related Question