CVL Ruling Database

<p>The CVL ruling dataset was synthetically generated to allow for comparing different ruling removal methods. It is based on the ICDAR 2013 Handwriting Segmentation database [1]. It was generated by synthetically adding four different ruling images resulting in a total of 600 test images. The pixel values are:</p> <ul> <li>255 background</li> <li>155 ruling</li> <li>100 text</li> <li>0 ruling and text (overlaping)</li> </ul> <p>For processing, a binary image must be generated which sets all pixels to 0 that are not 255. When evaluating, the line GT image can be found by setting all pixel having value 155 to one (e.g. linImg = img == 155). The text GT image can be extracted by setting all values below 155 to zero (e.g. txtImg = img &lt; 155). Then, true positives (tp), false positives (fp) and false negatives (fn) are defined as:</p> <ul> <li>tp = result &amp; linImg &amp; !txtImg</li> <li>fp = result &amp; !txtImg</li> <li>fn = !result &amp; linImg &amp; !txtImg</li> </ul> <p>The database ships with a&nbsp;Matlab that gives evaluation results if all images are already processed.</p>