Edge Detection
Detect edges in images (Sobel). Upload or drag-and-drop. Max 12 MB. Processing in your browser only. Large images may be downscaled for speed.
Detect edges in images. Upload or drag-and-drop an image (max 12 MB). The tool runs edge detection (Sobel) in your browser and shows a grayscale result. You can adjust strength. Large images may be processed at lower resolution for speed. Use Download to save the result.
Loading…
Examples
- Photo → edge detection → download outline
- Document → see edges
FAQ
Why is SVG not supported?
SVG is vector-based. Edge detection is applied to raster images (PNG, JPEG, WebP) using the canvas.
What is the maximum file size?
12 MB per image. Very large images may be processed at reduced resolution so they finish in a few seconds.
Is my image sent to a server?
No. Edge detection runs in your browser. Your image never leaves your device.
What algorithm is used?
A Sobel-style edge detector is used to highlight boundaries in the image. You can adjust strength.
What is the Sobel operator?
The Sobel operator computes the gradient of image intensity at each pixel using two 3×3 convolution kernels — one for horizontal changes (Gx) and one for vertical (Gy). The edge magnitude is sqrt(Gx² + Gy²). Stronger gradients appear as brighter edges.
What is the difference between Sobel and Canny edge detection?
Sobel is simpler and faster — it produces thick, continuous edges with some noise. Canny uses Gaussian blur, Sobel for gradients, non-maximum suppression, and hysteresis thresholding to produce thin, clean, accurate edges. Canny requires more computation.
What are real use cases for edge detection?
Edge detection is used in: object recognition and segmentation (computer vision), autonomous vehicles (road and lane detection), medical imaging (organ boundaries), document scanning (page detection), and artistic effects (line-drawing style output).
Why does edge detection turn the image to grayscale?
Edge detection measures intensity gradients. By converting to grayscale first, the algorithm works with a single channel (luminance) instead of three separate R, G, B channels, which simplifies the math and produces cleaner results.
What does the threshold / strength parameter control?
The strength parameter scales the edge magnitudes. Higher strength amplifies weaker edges (more detail, more noise). Lower strength keeps only the strongest edges (cleaner result, less detail). Adjust based on how much fine detail you want.
Can I use edge detection to make a line drawing from a photo?
Yes. Edge detection on a high-contrast photo produces a sketch-like line drawing. For best results: increase contrast first, then apply edge detection. Inverting the result (black edges on white) gives a pencil-sketch style.