Image Extraction

ExtractPageImages returns every image drawn on a page, together with where and how large it appears:

ex, err := extractor.New(page)
if err != nil {
    return err
}

pageImages, err := ex.ExtractPageImages(nil)
if err != nil {
    return err
}

for _, img := range pageImages.Images {
    goImg, err := img.Image.ToGoImage()
    if err != nil {
        return err
    }
    // ... encode goImg as PNG or JPEG ...
}

Passing nil uses the defaults. ToGoImage converts to a standard library image.Image, which is what lets you hand it to png.Encode or jpeg.Encode.

Displayed size is not pixel size

Each ImageMark separates the two, and mixing them up is the usual source of confusion:

• Width and Height are the dimensions as displayed on the page, in PDF units.
• X and Y are the position in PDF coordinates, measured from the image’s lower left corner.
• Angle is the rotation in degrees, if any.

The pixel dimensions live on img.Image instead. A small logo scaled up across half a page has a large Width and a small pixel count, so use the pixel values when writing files and the display values when reasoning about layout.

Inline stencil masks

Inline stencil masks are skipped by default. Pass options to include them:

pageImages, err := ex.ExtractPageImages(&extractor.ImageExtractOptions{
    IncludeInlineStencilMasks: true,
})

These are one-bit masks used for stencilled drawing rather than pictures, so they are usually noise when what you want is the document’s images. Turn them on when you are auditing everything a page draws.

Run the example

The example walks every page, converts each image to a Go image, and writes the lot into a zip archive. extractImagesToArchive holds the extraction and encoding loop.

git clone https://github.com/unidoc/unipdf-examples.git
cd unipdf-examples/extract
go run pdf_extract_images.go input.pdf images.zip

If this is your first time using UniPDF, follow the getting started guide to create an API key and set up your development environment.