Hello,
I've downloaded the IMGlib and have studied the sources for the functions I need. All of our image data is in 16-bit format. The specific functions that I need to run as efficiently as possible on 16-bit data are:
void IMG_boundary(const unsigned char * restrict in_data, int rows, int cols, int * restrict out_coord, int * restrict out_gray)
and
void IMG_thr_le2min(const unsigned char * restrict in_data, unsigned char * restrict out_data, short cols, short rows, unsigned char threshold).
If I write a 16-bit version (input and output as WORD* or u16*), how can I ensure that the compiler optimizes it well?
Can I take the assembly versions of the functions and change them to work with 16-bit data?
Thanks,
Mechi