Python code for feature extraction from a signature image in order to perform signature verification

I have coded a fetaure extractor in Python for the purpose of offline verification of signature images. I split the image into 64 cells recursively before extracting the features Following are the features I have extracted:

• Number of black to white transitions for each of the 64 cells
• Aspect ratio for each cell
• Centroid of the image and of each cell
• Number of black cells in image
• Angle of each cell from its bottom left corner to the image centroid
• Normalized sixe for each cell (cell size divided by number of black cells)
• Normalized angle of each cell from centroid (sum of angles divided by number of black cells)

Finally, I dump all of these values into their respective text files.

Essentially, these features enable one to distinguish between genuine signatures and those that are not authentic.

Note: This is not a machine learning code, it's merely a feature extractor