I am facing the issue while reconstructing the image, when the difference between the 2 public index != 1. 

for example: 
S = 15 
f(x) = 3x+15 % 17 
x1=1, y1=1 
x2=2, y2=4 
x3=3, y3=7 

now if 1st and 3rd try to retrieve the secreat 
c1 = -3/-2 = 1.5 
c3 = -1/2 = -1.5 

S = (1.5 + -10.5) % 17 
  = -9 % 17 
  = 8%17 
  = 8 which does not match with 15 that we had share 

but if 1st and 2nd or 2nd and 3rd will try to retrieve the share, they will, since diff 
between the public index = 1 

1) so letc for 1st and 2nd 

c1 = -2/-1 = 2 
c2 = -1/1 = -1 

S = (2 - 4)%17 
  = -2 % 17 
  = 15 which match with the secreat. 

Has anybody has encountered the same issue? 

I don't see your exact problem, because I only took a quick glimpse at it.

However, we had an issue with the division not coming out to be an integer, so, it was losing precision and the final results weren't coming out.

for instance, in integer division,

if x = 1, y = 2

x/y = 0, even though we know that 1/2 = 0.5

because 0.5 is less than 1.  It truncates the mantissa/decimal portion and becomes 0. 

So, the values you're using in the public indices may be causing a problem like that. 

It's likely that your coefficients

c1 = -3/-2 = 1.5

c3 = -1/2 = -1.5 

are actually c1 = 1 and c3 = -1
 

This is just a guess.