Convert 8-bit grayscale image to 7-bit grayscale image

0
votes

I want to convert my original 8-bit depth gray scale image to a 7-bit depth gray scale image (pixel values should be in the range [0 127].

I use the below syntax, however pixels with value equal to 255 will change to 128 (it seems they are rounded after division, i.e. 127.5 changes to 128). How can I resolve this issue and keep my pixel values in the range [0 127] after division?

RGB = imread('camera_man128x128.png')% read 8-bit image
RGB = RGB*0.5; %change pixel value to be in range to 0~127
               %however pixels with value 255 change to 128.
2
imagematlabgrayscale
You cannot have 7-bit data type in MATLAB. You can rescale your data to 0-127 integers though. Do you want that? – Sardar Usama
@SardarUsama yes, that't right. I want to store values in unit8 but only I want to change data scale from 0~127. – VSB
Floor your data, or divide by the max in your data after division and the multiply by 127 – Adriaan
Please edit your question to include the information you provided in the last comment and remove the ambiguous term '7-bit' – Sardar Usama

2 Answers

2
votes

The easiest way to do it is to use bitshift:

RGB = bitshift(RGB, -1);

This shifts the bit pattern of each uint8 value one to the right, equivalent to multiplication by 2-1 (i.e. division by 2), such that 255 will become 127.

1
votes

Convert your data type to double and then multiply with 0.5. Use floor to round towards negative infinity and then convert back to uint8. 

RGB = uint8(floor(double(RGB)*0.5));