DaiAlaa / OMR-Image-Processing

OMR

• Optical music recognition system written in Python which converts sheet music into a readable text file.

Algorithms & Techniques

Binarization:

• local Otsu

Rotating:

• Hough transform

Find Line width and Line space:

• run length coding

Split lines:

• Projection

Remove lines for scanned images:

• Line Removal with Projection

Remove lines for hand written images:

• Transfer music symbols to another image with Line width and Line space

Calculate staff line places:

• Projection

Segmentation:

• find contours

Classification:

• NN

Extract features:

• HOG

Centre of circles detection:

• Hough Circles

Primitive detection:

• Morphological operations

Prerequisites

• Python3
• Numpy
• Skimage
• Open CV
• Sklearn

Running the tests

• Put your input images inside the input folder
• python main.py $path_of_input_folder $path_of_output_folder
• Output appers inside the output folder

Testing

Accuracy

NN Hog Accuracy detection

• 95.7%

Overall Accuracy for scanned images:

• about 98%

Contributors

• Dai Alaa
• Dina Alaa
• Mohamed Monsef
• Nerdeen Ahmad