First step, average digits in the full 60k training set based on their labels:
Next, extract average features, using k = 5 (ie 5*5 tiles) and average categorize threshold t = 0.8:
$ ./create-average-images.py 5 work-on-handwritten-digits/label-average-images/Producing these features:
corresponding sw file, that we need in phi-transform.
Here are the raw 2000 test images:
Next, we phi-transform the test images:
$ ./phi-transform-v2.py 5 work-on-handwritten-digits/test-imagesProducing these images:
$ ./image_directory_edge_enhance.py 20 work-on-handwritten-digits/phi-transformed-images-v2/