Is it always necessary to convert image to double type before further execution?

Hi,

I want to threshold an image. I can read the image and it is a byte type (UINT 8). I want to pass this image to another function as it is but it asks me to convert to double before I can pass to another function where threshold is performed via MEX MATLAB. How can I pass the image as the same data type? Thanks in advance.

Answer 1

[Walter Roberson]

We don't know. You will need to tell us which function you are trying to pass it to.

[Krishna]

Thanks.

Answer 2

[Image Analyst]

Well apparently you can't. It said so. It said it wanted double and rejected your call when you tried to pass in uint8(). I don't know why, but that's the way it is. It seems like it could just cast it to double internally, but, for whatever reason, it doesn't. So I guess you have to live with it. Either use double() if it can take whatever range you have, or use im2double() if it wants it in the 0-1 range.

[Krishna]

Yes, you are right. I need to remap the image back to 'uint8' for further analysis. Now I am having a problem, like, in reading certain pixel values of an image and copy them to another new image (like image cropping). Thanks.

[Image Analyst]

Well you don't necessarily need to go back to uint8 for analysis - it just depends on what you're doing. I can't help anymore unless you provide more details. For example I have no idea if you're using imcrop() or doing the cropping some other way. I have a copy and paste demo (attached) in case that helps.

[Walter Roberson]

There is a routine to convert double precision images into uint8(). Or you can just use

uint8(floor(255 * DoublePrecisionImage))

[Krishna]

Actually I have my own data types defined in my header files. The image is defined as byte arrays and has image pointer too. I want to read the images in matlab (in mex) and convert this image to my data type and work from there. I am trying to copy all the pixel values from matlab's image to my own datatype's array of order (no_of_bands * rows * cols) as shown below.

for(bands=0; bands < no_of_bands; bands++)

   {
       image = getData_Image(next_image, bands);  //To work on each band at a time (R-G-B).
    for(r=0; r < rows; r++)

    {
        for(c=0; c < cols; c++) 

            image[r][c]= matlab_image[r+1][c+1][bands+1]; //image[r][c] is my byte type array.

    }
}

My problem is mainly how to access and assign each element of the matlab image because as far as I have understood my problem is in the "matlab_image[r+1][c+1][bands+1];" part of the code. Hope you are getting what I mean and thanks for all your suggestions.

[Jan]

@Krishna: This is not Matlab. We do not know: getData_Image function, type and size of "image", value or "rows" and "cols", size and type of "matlab_image" and why you use 1-based indexing here. So currently there is no chance to guess, what you are doing.

[Krishna]

Sorry that it was not clear. Actually 'image' variable is byte-array type defined in my header file along with that 'getData_Image' function.1-based indexing is for the matlab arrays as compared to C which is 0-based since I want to transfer pixel values to corresponding places. If all these are unclear, can you please let me know how to access each value of image (which is obviously a matrix) we read in matlab and pass to a C type multidimensional array.