Help on converting dubble to uint8 ?

Hey,

I have wave signal that change between 2-20 and it's type of dubble. I need to convert it to type uint8, for example:

The wave signale is INDX(change from 2 to 20) type of dubble. Convert signal is: ADR = uint8(INDX);

the result of that conversion gives me values from 0 to 20. Even when ADR is runs from 0 to 18 or 1 to 19 I get the same result : ADR = 0 to 20 ?

Is it possible to get the same values of INDX after using INDX input ? for example, if INDX is running from 3 to 17, the result (ADR=uint8(INDX)) will be 3 to 17 instead 0f 0 to 20 ? or, if INDX is running from 2 to 18, the result (ADR=uint8(INDX)) will be 2 to 18 instead 0f 0 to 20 ? or, if INDX is running from 2 to 20, the result (ADR=uint8(INDX)) will be 2 to 20 instead 0f 0 to 20 ?

I need to keep the same values of the start and end(of INDX) after using uint8(INDX). uint8 type ,must be use.

Thanks, Henry

Answer 1

[Walter Roberson]

You must have made a mistake somewhere.

>> uint8(3:17)
ans =
  3    4    5    6    7    8    9   10   11   12   13   14   15   16   17

In the expression 3:17 the values are double precision values, so this shows that converting double precision values to uint8 works as expected for small non-negative values.

[Henry Buck]

Hi,

well, that's the problem: the values of INDX are not a vector as you wrote.

INDX = round(A/B). B is a constant - let say 40. A is kind a sine wave - max value is 800 and min value is 0 - that gives INDX is between 0 to 20 In other case it couild be 90 to 700 - that gives INDX between 2 to 18..

whent I am using uint8(INDX), it gave me 21 values - 0 to 20.

Why is that ?

thanks, Henry

[Stephen23]

@Henry Buck: please show us exactly the 21 values with this behavior, and how you are converting them to uint8. Then we can explain what is happening.

[Image Analyst]

Is it possible you use im2double() somewhere along the way? That might rescale and shift things. gray2mat() also scales and shifts to the 0 to 1 range. Also, uint8 rounds internally. It does not just chop off the fractional part.

Like everyone else is saying, you need to show more code, or at least a small similar example, like Walter did, that proves what you're saying.

[Walter Roberson]

Not being a vector is irrelevant.

>> A = 53*(reshape(3:17, 3, 5) + rand(3,5)); B = 48; INDX = uint8(A/B)
INDX =
    4    8   10   14   17
    5    9   12   15   19
    6    9   13   16   19

[Henry Buck]

hey guys,

I have to be honest. It is my mistake.

Thanks again.

Henry