Here is the code that I came up with, but its probably very inefficient. Are there any ways to make this more efficient?

char ScanKeySingle(void)
{
unsigned int key16;
char key8 = 0;
int i;
unsigned int mask = 0x8000;
bit multiple = 0;

// scan and produce 16-bit key
key16 = ScanKey();

// if it was zero then noise
if(key16 == 0)
return 0x80;

// check each bit, keeping track of the
// 4-bit encoded value and multiple
// keypresses
for(i=0; i<16; i++)
{
if(mask & ~key16)
{
if(multiple)
return 0x10;
key8 += i;
multiple = 1;
}
mask = mask >> 1;
}

return key8;
}