Clarification for baseline

[fgregg]

In the explanation of the baseline property, the spec says that "The polynomial is in the coordinate system of the line, with the bottom left of the bounding box as the origin."

In the example, the polynomial is y = 0.015x - 18

So, at left edge of the bounding box, y should equal -18 relative to the bottom, left corner, and this is what the text says. However the picture has the baseline starting at +18 on the left edge.

Further, I don't see how it's logically possible for the intercept to be a negative number since that would mean that the baseline started outside the bounding box. But the bounding box is supposed to always contain the baseline.

[kba]

Good points, I'd be happy to include clarifications.

@tmbdev @zuphilip ?

[zuphilip]

The example comes from Tesseract wiki pages: https://github.com/tesseract-ocr/tesseract/wiki/FAQ#how-to-interpret-hocr-baseline-output

[marschap]

Negative values for the last parameter for baseline are correct (even necessary):

• hOCR's Y coordinates start with 0 at the top, and increase downward (see red coordinate arrows in the example)
• The last parameter for baseline is taken as an offset from the lower border of the bbox With Y coordinates growing downwards, the lower border of the bbox has higher numeric values
• To be inside the bbox, the last parameter for baseline needs to be negative

[marschap]

My issues with the baseline property values are with the grammar:

• It mentions property-value = float int I.e. exactly one float followed by exactly one int
• Yet in the the description bewlow, it talks about a polynomial of degree n, with n>=1
• Accepting this, the grammar needs to read property-value = float+ int to be consistent with the description.

[marschap]

Sorry got the BNF wrong: property-value = +(float) nint This also enforces the "final parametel must be negative" topic