Image Deskew Tool

Automatically detect and correct tilted scans, documents, and crooked photos.

Tested tool guide Tested browser tools Checked August 16, 2026

What Image Deskew Tool does and how it behaves

Image Deskew Tool is for an image whose page, text baselines, table rules, or architectural edges lean by a small angle even though the subject otherwise faces the viewer. It detects the dominant tilt and applies the opposite rotation so those cues become level or upright. The common surprise is that deskewing is not perspective correction. If one side of a photographed page appears larger than the other, a single rotation cannot make all four edges rectangular. Rotation can also create empty corner areas or require some edge trimming.

How the result is produced

Detecting the correction angle

Skew is the consistent angular offset shared by features that should be horizontal or vertical. The tool evaluates the selected image for a dominant alignment, chooses a correction angle, and rotates in the reverse direction. Text rows, ruled forms, page borders, and long straight edges usually provide clearer evidence than isolated objects or irregular artwork.

Framing the rotated image

A rotated rectangle no longer matches its original axis-aligned bounds. Keeping every source pixel therefore requires blank triangular regions around parts of the perimeter, while filling a clean rectangle requires cropping, scaling the rotated image until it covers the frame, or synthesizing missing corner content. Check the corrected result around signatures, page numbers, stamps, and corners, where a small crop can remove meaningful material.

Good uses

Straightening a flatbed scan made with a contract or application form resting slightly crooked against the scanner edge.
Correcting a phone photo of a receipt, whiteboard, poster, or sign whose horizontal lines all slope in the same direction.
Preparing a scanned page for OCR when tilted text rows could interfere with line segmentation or make the recognized layout harder to follow.

Limits and checks

A trapezoidal page is not merely skewed. When opposite edges are not parallel or vertical edges converge, one correction angle may level one feature while leaving others tilted. The image can be correctly deskewed and still need perspective correction.
Automatic detection can be ambiguous when the image has little text, no long edges, several competing angles, a decorative border, or a deliberately diagonal composition. Inspect whether genuinely horizontal or vertical subject features are level after correction.
Deskewing changes pixel positions through rotation. Fine text can become slightly softer, and the rectangular result may contain blank corners or omit perimeter content. Retain the original image when marginal notes, seals, or exact source pixels matter.

Common questions

Will this fix perspective distortion or a curved book page?

No, not when distortion causes parallel page edges to converge or makes one region larger than another. Deskewing applies one angular correction across the image. Perspective distortion needs corner or perspective correction, while curved book pages need dewarping. Deskew can still remove a remaining overall tilt after those distortions are corrected.

Will deskewing keep the original dimensions and every pixel?

Not necessarily. Rotation changes where the image corners fall relative to a rectangular output. A result that preserves all source content may include empty margins; a result that fills the frame may crop the perimeter. Rotation also resamples pixels, so repeated deskewing and saving can slightly soften fine text. Keep the unmodified original if edge content matters.

References and verification

The behavioral notes were checked against the browser implementation. Standards and primary references below define the relevant format, formula, or platform behavior.