PDF Image Alternatives
What It Is
In a tagged PDF, an image's text alternative lives in the /Alt entry of the /Figure structure element in the tag tree. When a screen reader encounters a /Figure, it announces the contents of /Alt; if /Alt is missing, it announces "Figure" or "graphic" and nothing else -- the same failure mode as an <img> with no alt attribute in HTML. For glyphs whose visual form should be read as different text (a decorative drop cap, a stylised ligature, an image of a mathematical symbol), the /ActualText entry replaces what the screen reader speaks. Purely decorative images must be tagged as /Artifact so they are excluded from the reading order entirely, analogous to alt="" on HTML <img>[1].
Why It Matters
The failure is mechanistic, not rare. A Word document with an image that has no alt text in the source exports to PDF as a /Figure tag with no /Alt entry; the screen reader announces "Figure" and moves on, and the reader has no way to recover the content. A decorative divider image that was never artifacted produces a "Figure" announcement between every section, breaking reading flow without conveying anything. A chart exported as a single flattened image with a one-word /Alt ("Chart") technically passes a tag-presence check and still leaves a blind reader with no access to the data. And a scanned report with no OCR layer is, to a screen reader, a sequence of image pages containing zero text -- there are no glyphs to tag in the first place.
How It Relates to WCAG
This is the PDF binding of WCAG 1.1.1 Non-text Content[2]. EN 301 549 clause 10.1.1.1 applies the same requirement to non-web documents, and ISO 14289-1 (PDF/UA-1) specifies the concrete mechanism -- /Alt on /Figure, /ActualText for replacement glyphs, and /Artifact marking for decoration[3][4][1].
Practical Implications
- In Word, right-click the image and choose Edit Alt Text to write a description, or check Mark as decorative so the export writes the image as an artifact. The source-document metadata is what the PDF export reads -- fixing alt text in the PDF after the fact is strictly rework.
- In InDesign, set Object Export Options > Alt Text per image (or pull from an XMP field) so the export to tagged PDF carries
/Altthrough. - In Acrobat Pro, use the Reading Order tool to edit alternate text on existing
/Figuretags, or open the Tags panel and change a decorative/Figureto/Artifactso it drops out of the reading order. - Run OCR before tagging any scanned document -- there is nothing to describe until the glyphs exist as text.
- For complex charts and infographics, write a short
/Altthat names the chart and give the detail in body text or an appendix the reader can navigate to. A long/Altburied on a/Figureis not navigable. - Use
/ActualTextonly for the narrow case of glyphs whose visual form should be spoken as different text (drop caps, scientific notation rendered as images, stylised ligatures). Do not use it as a substitute for/Alton informative images. - A nearby visible caption does not substitute for
/Alt. The screen reader announces the/Figureindependently of surrounding text; without/Altit still announces "Figure" and nothing else.