You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
First step in the skew detection is thresholding: converting the input image to 1bit black and white. . The trouble is determining the right threshold value - pixels with lower value are considered black
and above are white. As a default, Deskew calculates the threshold automatically (Otsu's method) which works well enough. However, when there are e.g. large dark areas (scanner >> page) the image histogram is affected and the calculated threshold is not ideal for the text.
Content rectangle aka "area of interest" can already be passed to Deskew and auto thresholding
should only take this area into account.
The text was updated successfully, but these errors were encountered:
First step in the skew detection is thresholding: converting the input image to 1bit black and white. . The trouble is determining the right threshold value - pixels with lower value are considered black
and above are white. As a default, Deskew calculates the threshold automatically (Otsu's method) which works well enough. However, when there are e.g. large dark areas (scanner >> page) the image histogram is affected and the calculated threshold is not ideal for the text.
Content rectangle aka "area of interest" can already be passed to Deskew and auto thresholding
should only take this area into account.
The text was updated successfully, but these errors were encountered: