skewScore in scandata is corrupt

Bug #203383 reported by raj
2
Affects Status Importance Assigned to Milestone
Scribe2
Confirmed
Medium
Unassigned

Bug Description

We think the code in the scribe image loop that writes skewScore to scandata.xml is writing bad data.

Example in item socialismvscivil00bras

The skewScore of leaf 272 has a > in it:

    <page leafNum="272">
      <handSide>LEFT</handSide>
      <pageType>Normal</pageType>
      <addToAccessFormats>true</addToAccessFormats>
      <rotateDegree>-90</rotateDegree>

      <skewAngle>-0.25</skewAngle>
      <skewAngleDetect>-0.25</skewAngleDetect>
      <skewScore>&gt;3.06</skewScore>
      <skewActive>true</skewActive>
      <origWidth>2912</origWidth>
      <origHeight>4368</origHeight>

      <cropBox>
        <x>987</x>
        <y>1198</y>
        <w>1814</w>
        <h>2718</h>
      </cropBox>
      <ppi/>

      <pageNumber>246</pageNumber>
    </page>

Tags: imaging
siznax (siznax)
Changed in scribe2:
assignee: nobody → steve-archive
status: New → Confirmed
siznax (siznax)
Changed in scribe2:
importance: Undecided → High
Revision history for this message
Hank Bromley (hank-archive) wrote :

Here's another example of bad scandata that may have the same cause.

For some fields, these are now flagged during ProcessJP2 - see redrow category "ProcessJP2 bad scandata (non-numeric rotation)." This one is from book universalantholo01garn, which shows "Rotation e>90 is not numeric" in its log. Here's the scandata for the page (the bad element is rotateDegree):

<page leafNum="379">
  <handSide>RIGHT</handSide>
  <pageType>Normal</pageType>
  <addToAccessFormats>true</addToAccessFormats>
  <rotateDegree>e>90</rotateDegree>
  <skewAngle>0.55</skewAngle>
  <skewAngleDetect>0.55</skewAngleDetect>
  <skewScore>3.76</skewScore>
  <skewActive>true</skewActive>
  <origWidth>2912</origWidth>
  <origHeight>4368</origHeight>
  <cropBox>
    <x>137</x>
    <y>357</y>
    <w>2304</w>
    <h>3556</h>
  </cropBox>
  <ppi/>
  <pageNumber>361</pageNumber>
</page>

Revision history for this message
Hank Bromley (hank-archive) wrote :

Another case of bad scandata. This is a scribe2 book scanned in January, without compress-in-the-field.

id: johnsonianaorsup00crok

from the scandata for leaf number 388:

  <rotateDegree>-9</rotateDegree>

(This one wasn't flagged as bad scandata; it redrowed because the missing final 0 caused the image to be mis-rotated by roughly 90 degrees, which led to a failed crop operation, as the crop box didn't fit with the image turned sideways.)

raj (raj-archive)
Changed in scribe2:
importance: High → Medium
siznax (siznax)
Changed in scribe2:
assignee: siznax (steve-archive) → nobody
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.