swtloc 1.0.0.0

Python Library for Stroke Width Transform

SWTloc : Stroke Width Transform Text Localizer

Description

This repo contains a python implementation structured as a python package pertaining to the text localization method as in a natural image as outlayed in the Research Paper :-

Detecting Text in Natural Scenes with Stroke Width Transform. Boris Epshtein, Eyal Ofek & Yonatan Wexler (June, 2010)

Installation

---

Installation

pip install swtloc

Usage

---

• Stroke Width Transform

  This function applies Stroke Width Transform on a list of images or a single image, and once done with the transformation the results are stored in the directory mentioned save_rootpath under the name of the image being processed (Process images with different names)

  from swtloc import SWTLocalizer
  
  swtl = SWTLocalizer()
  imgpaths = ... # Image paths, can be one image or more than one
  swtl.swttransform(imgpaths=imgpath, save_results=True, save_rootpath='swtres/',
                    edge_func = 'ac', ac_sigma = 1.0, text_mode = 'wb_bf',
                    gs_blurr=True, blurr_kernel = (5,5), minrsw = 3, 
                    maxCC_comppx = 10000, maxrsw = 200, max_angledev = np.pi/6, 
                    acceptCC_aspectratio = 5)
  

  

  ---

• Individual & Grouped Bounding Boxes and Annotations

  These group of functions returns bounding boxes to individual components and grouped bounding boxes of words.

  • Minimum CC Bounding Boxes

    Generate Minimum Bounding box for each of the Connected Components after applying SWT, can be a rotated rectangle.

    min_bboxes, min_bbox_annotated = swtl.get_min_bbox(show=True, padding=10)
    

    

  • Extreme CC Bounding Boxes

    Generate External Bounding box for each of the Connected Components after applying SWT.

    min_bboxes, min_bbox_annotated = swtl.get_extreme_bbox(show=True, padding=10)
    

    

  • CC Outline

    Generate Outline of each of the Connected Components after applying SWT.

    comp_outlines, comp_outline_annot = swtl.get_comp_outline(show=True, padding=10)
    

    

  • CC Bubble Bounding Boxes

    Generate Bubble Bounding box for the grouped letters into a word.

    respacket = swtl.get_grouped(lookup_radii_multiplier=1, sw_ratio=2,
                         cl_deviat=[13,13,13], ht_ratio=2, 
                         ar_ratio=3, ang_deviat=30)
    
    grouped_labels = respacket[0]
    grouped_bubblebbox = respacket[1]
    grouped_annot_bubble = respacket[2]
    grouped_annot = respacket[3]
    maskviz = respacket[4]
    maskcomb  = respacket[5]
    

    

  *CC = Connected Component

Documentation

---