While preparing some images for OCR, I usually discard those with low DPI, but as this happens often I thought to try some image processing and on suggestion (<span style="color:rgb(17,17,17);font-family:&#39;Helvetica Neue&#39;,Arial,sans-serif;font-size:14px;line-height:19px;text-align:left;background-color:rgb(253,253,253)">morphological</span> operations) I tried ndimage.morph with idea to play around binary_dilation<div>
<br></div><div>Images were G4 TIFFs which PIL/MPL can&#39;t decode, so I convert to 1bit PNG which I normalized after to 0 and 1.</div><div><br></div><div>On sample img I applied:</div><div><br></div><div>ndi.morphology.binary_dilation(img).astype(img.dtype)</div>
<div><br></div><div>and</div><div><br></div><div>ndi.morphology.binary_erosion(img).astype(img.dtype)</div><div><br></div><div>I attached result images, and wanted to ask two question:</div><div><br></div><div>1. Is this result correct? From what I read today seems like what dilation does is erosion and vice versa, but I probably overlooked something</div>
<div>2. Does someone maybe know of better approach for enhancing original sample for OCR (except thresholding, for which I&#39;m aware)?</div><div><br></div><div>TIA</div><div><br></div><div><img src="cid:ii_1366991fd2b11631" alt="Inline image 1"><br>
</div>