How to Extract Images from a PDF: 4 Methods at Full Quality

By Michael Peters · Published July 28, 2026 · 7 min read

← Back to Blog

Why Extract Images from a PDF?

The classic scenario: the brochure exists, the report is published, the manual is finished — and the original image files are gone. The photographer’s originals were never archived, the designer has moved on, the source InDesign package is on a backup nobody can find. The PDF becomes the only surviving container for photographs, product shots, logos, and diagrams that someone now needs for a website, a presentation, or the next edition.

The good news is that a PDF usually stores images as complete, self-contained files at their original resolution. Extract them properly and you get the actual embedded data back — not a lossy copy. Extract them badly (by taking screenshots) and you throw that quality away. This guide covers doing it properly.

How Images Live Inside a PDF

Understanding a little of the internals explains almost every surprise you’ll meet later. A PDF stores each raster image as an image XObject — a named, reusable resource containing the pixel data, usually keeping its original compression: JPEG data is stored as JPEG (DCTDecode), scans are often fax-compressed (CCITTFaxDecode or JBIG2Decode), and other images use lossless Flate compression. The page content stream then draws the XObject at some position and scale. (Our guide to PDF file structure covers how these pieces fit together.)

Three consequences worth knowing before you extract:

• A repeated image is stored once. A logo that appears on 50 pages is one XObject drawn 50 times — extraction gives you one file, which is what you want.
• Vector graphics are not images. Charts, diagrams, and logos exported from Illustrator are drawing instructions, not pixels. Image extraction will not return them — you need to render the page instead (see below).
• The drawn size is not the stored size. An image displayed at 5 cm wide may be a 3,000-pixel-wide original squeezed down — extraction recovers the full 3,000 pixels. This is exactly why extraction beats screenshots.

Method 1 — Adobe Acrobat Pro (Export All Images)

Acrobat Pro has a dedicated bulk-extraction feature that most users never find. Open the PDF and choose Tools > Export PDF > Image, pick a format (JPEG, JPEG 2000, TIFF, or PNG) — then tick the Export all images checkbox. That checkbox is the crucial step: with it ticked, Acrobat saves every embedded image as its own file instead of rendering whole pages.

Click the settings icon next to the format to control compression quality and colour management, and — genuinely useful on real documents — to exclude images below a minimum size, which filters out icons, bullets, and decorative rules that would otherwise litter the output folder.

For a single image, two quicker routes: right-click the image and choose Copy Image, then paste into an image editor; or select Edit PDF, right-click the image, and use Save As Image. Avoid the Snapshot tool for anything you care about — it captures at screen resolution, not the stored resolution.

Note that this is a Pro feature: the free Acrobat Reader cannot export images.

Method 2 — Free Online Extraction

For a quick job without installing anything, Mapsoft’s PDF Hub Extract Images tool is free: upload a PDF and it finds every embedded image and returns them in a ZIP. Each image is extracted at its original stored resolution as a lossless PNG, so nothing is recompressed on the way out, and an optional page range limits extraction to the section you need. Files are processed on Mapsoft’s own UK servers and deleted automatically within 24 hours.

The usual online-tool caveat applies: the document leaves your machine. For confidential material, use a local method — Acrobat, the command line below, or a desktop application such as Mapsoft’s PdfHub Local, which never uploads files.

Method 3 — The Command Line (pdfimages)

The free pdfimages utility from the Poppler project (installable on Linux, macOS via Homebrew, and Windows) is the professional’s tool for this job, and it has one feature nothing else here offers: an inventory. Run

pdfimages -list report.pdf

and you get a table of every image in the file — page, dimensions, colour space, bits per component, encoding, and effective DPI — before extracting anything. It is the fastest way to answer “what’s actually in this PDF, and at what quality?”. Then extract with

pdfimages -all report.pdf out/img

The -all flag writes each image in its native format — JPEGs come out as the original .jpg files, byte for byte — which is the gold standard for lossless recovery. Two quirks to expect: transparency (soft) masks are stored as separate grayscale images and extract as separate files, and fax-compressed scans may extract in formats you’ll want to convert afterwards.

Method 4 — Programmatic Extraction

For batch pipelines — recovering assets from an archive, feeding figures into a machine-learning dataset — script it.

Python. PyMuPDF extracts each image’s original bytes with its native extension in a few lines: iterate page.get_images(), call doc.extract_image(xref), and write the returned bytes. pikepdf offers lower-level access to the raw streams when you need to handle unusual encodings or masks yourself.

Adobe PDF Services API. The cloud Extract API returns a structured JSON of the document’s contents with each figure saved as a separate file — heavier than simple extraction, but the right shape when you also need to know where each image sat in the document structure.

.NET and Java. Libraries such as Aspose.PDF and iText expose embedded images as collections you can enumerate and save — suitable for server-side processing on your own infrastructure.

Scanned PDFs: One Big Image per Page

Extract images from a scanned PDF and you will get exactly one image per page — the full page, text and all. That is not a tool failure: a scanner photographs the whole sheet, so the whole sheet is the image. If you need an individual photograph from within a scanned page, extract the page image at full resolution and crop it in an image editor. And if what you actually want from the scan is the text, that is an OCR job, not an extraction job — see our guide to OCR in Adobe Acrobat.

Extracting vs Converting: Pick the Right Operation

These two jobs are constantly confused, and picking the wrong one wastes quality or effort:

• Extraction pulls the original embedded images out at their stored resolution. Use it to recover photos, logos, and graphics at maximum quality. Vector artwork and text are not included.
• Conversion (rendering) rasterises each whole page to a new image at a DPI you choose — capturing everything, including vector charts and text. Use it for page thumbnails, previews, slides, or to capture vector diagrams as pixels. Our PDF-to-images guide covers this side.

Rule of thumb: a photo inside the document → extract; the page itself as a picture → convert. And don’t confuse either with embedded file attachments — files carried inside a PDF as a portfolio or annex are a different mechanism, covered in our guide to extracting embedded files.

When the Results Surprise You

• Fewer images than expected. Reused images extract once; vector graphics don’t extract at all; some tools skip small inline images embedded directly in the page stream.
• More images than expected. Layout applications sometimes slice one visual into several tiles (or store the image and its transparency mask separately), so a single picture on the page can arrive as multiple files.
• Colours look wrong. Print PDFs often carry CMYK or ICC-profiled images; naive extraction can shift colours when saving to RGB formats. If colour fidelity matters, extract with a tool that handles colour management (Acrobat with colour settings, or pdfimages -all keeping native data) and convert deliberately.
• Image is rotated or flipped. The page may draw the image with a transform; the stored data extracts in its original orientation. Rotate it back in an image editor.
• Quality is worse than the page suggests. The creating application may have downsampled images when the PDF was made (a common export setting). Extraction faithfully returns what is stored — pdfimages -list tells you the truth about what that is.

One final note: extracting an image does not change who owns it. Photographs and illustrations in a PDF are typically copyrighted — make sure you have the rights before reusing them.

Tool Comparison: Which Method When