Specularity removal in multi-viewpoint images

The package is an open source utility to remove specularity in Images captured from multiple viewpoints.

Disclaimer

The method uses Homography to find the best pixels available in other viewpoint images which are specular-free. It works only for planar and textured objects.

Requirements:

• Numpy
• OpenCV

Works with both Python 2.7+ and Python 3+

Usage

The package is pip installable. The easiest way to install is by using pip.

$pip install specularity_removal

You can also download the git repo and install by running the below command.

$python setup.py

To use the specularity removal module in your code, follow the below sample snippet. It accepts an array of image path names. Some constraints are,

• It should be in same order as the object is in real.
• All the images should be of equal resolution.
• Images should have enough textures.
• It is faster if you pass the downscaled version of actual images.

import specularity_removal as sp

fnames = [
    'data/pshycology-book-1.jpg',
    'data/pshycology-book-2.jpg',
    'data/pshycology-book-3.jpg'
]
sp.remove_specularity(fnames)

Example 1

Using two input images

Output

Example 2

Using three input images

Output