OpenCV Python script to collect a sample of handwritten characters from an (24x16) grid image

The objective of this program is to extract cells from a grid and resize into (i.e 28x28) pixel images. It is important to correct orientation and angle's of an image as it may be skewed and not perfect. Each row represents a class of an image and there are total of 16 columns in a row. Row 1 to 24 is on the first image, and row 25 to 44 is on the second image. Input images are numbered sequentially, therefore every two consecutive images can be considered as a set. So, there will be 704 (44x16) images in each set, 16 images per class. It is important to clean/exclude border lines when a character is touching the lines. Once the project is accepted I will provide additional images of about 230 for testing. Program should have good results on these test images.