sid-kumar-iyer/internship_at_iitk

Research internship @ IITK during summer 2018

internship_at_iitk's Issues

Sanity of input coordinates for area interpolation

When interpolating the pixel intensity based on area, your code does not check for the sanity of input. You may either:

1. Sanitize your code, or
2. replace it with:

import numpy as np
import math

def spatial_interpolation(image, p) :
  ## sanity check
  if (np.any(np.array(p) < 0) or
      np.any(np.array(p[::-1]) >= image.shape)) :
    raise IndexError('%s: [LB: 0 UB): %s', p[::-1], image.shape)

  x, y = p
  dx, dy = p - np.floor(p)

  # lg.debug((x, y, dx, dy))

  I = np.array([[
    image[int(math.floor(y)), int(math.floor(x))],
    image[int(math.floor(y)), int(math.floor(1 + x))]
  ],[
    image[int(math.floor(1 + y)), int(math.floor(x))],
    image[int(math.floor(1 + y)), int(math.floor(1 + x))]
  ]])

  A = np.matmul(
    np.array([[1 - dy, dy]]).T,
    np.array([[1 - dx, dx]])
  )

  value = np.sum(I * A)

  # lg.debug((I, A, value))

  return value

This piece of code raises exception... The caller is responsible to handle the exception

Limit of 1000 on number of lines or something like that...

internship_at_IITK/snapping.py

Line 31 in 57e53b6

vis1=np.full((1000,),1)

internship_at_IITK/snapping.py

Line 217 in 57e53b6

vis=np.full((1000,),1)

Instead of hardcoding the memory allocation, you may:

1. wait for the limit to be known before you initialize the array,
2. try-catch to update the limit when the program crosses the limit.
3. Heuristically, set a larger memory --- strongly suggested against, because, it limits the capacity of parallel of the processing