Document photo scanning correction
Adjust the four corners of photos to straighten the document. Supports automatic selection for bright paper and saving multiple PNG·PDF files.
Document photo
Photo · file name · coordinates are processed only in browser. Input and results are cleared on refresh or page navigation.
Adjusted document
After checking the corners, click Adjust and PDF Generate.
The original image is not modified. It is composed of a PDF image and does not include searchable text or OCR. When sharing, do not include images, document content, or file names.
Share result
Share the tool link and result text together. No URL is included in the result. Check the below content and then share it.
Method for document photo scanning correction
Input example
640×480px In an example, the auto-searches bright paper on a dark background and checks the four corners. Selecting A4 vertical ratio expands the document into a 210:297 rectangle. For an 2 page example, it becomes a 2 page PDF in order. Recreating it after moving left or right updates the new order.
By specifying the four points at top-left, right-top, right-bottom, and bottom-left, the document is flattened into a rectangle via perspective transformation, and neighboring pixels are interpolated. Overlaps or intersections are indicated as errors. The default is the entire image; select the paper area directly.
Auto find is a limited search that identifies the largest bright area on a dark background. It fails or misidentifies in bright backgrounds, shadows, color paper, multiple documents, or large rotations. Always check the corners after the search. If failed, retain the existing coordinates and adjust manually.
The estimated ratio based on the width is not guaranteed to match the actual print ratio. For an actual A4·Letter vertical document, please select this ratio. No function to restore curved surfaces, lens distortion, or blurred text is available. The result is scaled down within the specified long width without zooming, and transparent pixels are treated as white. The total output is 24 million pixels, a total image size of 75MB, and a PDF size of 100MB or less.
After selecting a page, moving forward or backward changes the PDF order. Each page can be saved as PNG, and the entire document as a PDF. Photography details and file names are not copied to the new file. HEIC·WEBP·movable PNG are not supported.
Frequently Asked Questions
Can automatic corner detection be applied to all photos?
No. It is a limited method that finds the largest bright paper on a dark background. It may fail on bright backgrounds, shadows, colored paper, or multiple documents. Directly adjust the four points or use numerical input, and check the edges after searching.
Is the document content sent to the server or searched in PDF?
Photo · coordinates · document content is processed only in browser and not saved. PDFs contain adjusted images and do not include OCR or searchable text. Input and results are cleared on page navigation.
How is the actual ratio of paper adjusted?
Only the length of the photo cannot accurately determine the actual paper ratio. Choose the ratio for A4 or Letter vertical documents. Use the default estimate for documents with different ratios. Curved paper or lens distortion cannot be restored by this flat correction.
Calculation criteria and reference materials
- OpenCV — planar perspective transformation
Concept of perspective correction on a plane using four points. This tool uses its own interpolation and does not load OpenCV.
- MDN — Canvas pixel reading
Pixels and color space of the image read by the browser.
- MDN — Canvas image saving
Encoding browser pixels as a new image file
Verification of calculation criteria: · Calculation·verification principles · Report errors