Pattern Recognition 1 - Expert System¶

Objectives

Create a working image analysis and pattern recognition pipeline
Object labelisation and feature extraction
Creating an Expert System for Optical Character Recognition (OCR)

Libraries needed for this lab

Numpy
Matplotlib
Scikit-image

(all included in the Anaconda distribution)

Object detection¶

Given the following image :

In [1]:

Copied!





import numpy as np
from matplotlib import pyplot as plt
from skimage.io import imread
%matplotlib inline

im = imread('../data/doc1.png')
print(im.shape, im.dtype)

plt.figure(figsize=(20,15))
plt.imshow(im, cmap=plt.cm.gray)
plt.show()
import numpy as np
from matplotlib import pyplot as plt
from skimage.io import imread
%matplotlib inline

im = imread('../data/doc1.png')
print(im.shape, im.dtype)

plt.figure(figsize=(20,15))
plt.imshow(im, cmap=plt.cm.gray)
plt.show()

(851, 1068) uint8

Segment the image to isolate the text from the background
Label individual characters and extract the centroid position, the bounding box, and useful features for each character.

Useful documentation: skimage.measure module.

In [2]:

Copied!

## -- Your code here -- ##

## -- Your code here -- ##

Line detection and letter ordering¶

Find the labels (objects) that belong to each text line
Order each character in a text line from left to right

In [3]:

Copied!

## -- Your code here -- ##

## -- Your code here -- ##

Expert System classification¶

Using the features available with the regionprops method of scikit-image, propose a method to automatically recognize some letters.

In [4]:

Copied!

## -- Your code here -- ##

## -- Your code here -- ##

Full OCR pipeline¶

Using all the previous exercices, create a method that takes as input a text image, and outputs the recognized text.

In [5]:

Copied!

def OCR(im):
    text = ''
    
    return text

print(OCR(im))
def OCR(im):
    text = ''
    
    return text

print(OCR(im))