A small collection of functions for various photography tasks. The main one is the split_multi_scanned_photos
that I needed to crop whole flatbed scans on my Epson V600, their software is crappy and took too long to select and scan each photo so I would scan a grid of 4 photos and split them using this function. Any PR are welcome as well and requests for new functions.
- matplotlib
- tifffile
- scikit-image
- tqdm
This is not yet on pypi so you need to install it manually.
- Clone the repository to your local hard drive
- open up anaconda prompt and change directory to
pip install -e .
- Start using the library scripts! `from split_multi_scanned_photos import split_multi_scanned_photos1
- Optionally you can run the script
split_multi_scanned_photos
by itself. That is what I do to debug scans that aren't cropping properly
I've included a main script that I use for my processing as an example. The small script grabs all tiffs found in a specified folder and tries to split them. I scanned about 40GB of photos so far and it has worked fairly well on ~95% of the scans, it runs into problems if the image has large areas of white that may be confused for background, at that point run the function in debug mode and only on that image to see why it may be failing.
I used an Epson V600 in advanced mode set to 800dpi and 48 bit depth. This produced whole flatbed scans of about 350MB per flat bed scan, after cropping images were ~77MB. I scanned the images as tiff's for lossless quality. I've processed about 200GB of photos and ~95% of them are well cropped, the rest may need to be manually cropped. It also may fail if large parts of the image are white and therefore close to the color of the flatbed background.
split_multi_scanned_photos
Note: Function will take the filename of the original scanned image and append the image index to the new filename.
- path_im : path to image on hdd.
- path_output : path to where you want images to be save to, otherwise an output directory will be created at the root of the files. The default is None.
- region_threshold :regions of interest smaller than these size will be ignored . The default is 1e6.
- pad : pads the bounding box to prevent accidental cropping of photos edge. The default is 50px.
- deskew : Deskew photo on the fly. The default is True.
- debug : Show intermedaite images for debugging. This will NOT save images to the hard drive. The default is False.
Other Notes:
region_threshold
really depends on the size of your images. This is used to remove small regions in segmentation smaller than the size of an imagepad
adds a buffer area so that you aren't accidentally cropping the original image, this does mean it will leave a small blank margi which I am working on a function to remove.deskew
uses skimage.transform.rotate function with default settings which is set to fill bg with maximum itensity color of original image (some hue of white), the original range is preserved and interpolation is set to nearest neighbor as is the default for skimage
thumbnail generator
- This function will scale all the photos in a folder by a specified amount. I used this to batch downsaple the photos to share with family/friends.