Home▸Algorithms & AI▸Corners, Blobs and Keypoints: Harris, SIFT & ORB Lab

🧭 Corners, Blobs and Keypoints

Two synthetic photos of the same scene, one rotated, scaled and made noisy, with live Harris corner, SIFT blob and ORB keypoint detection so you can see which keypoints survive and get matched across the viewpoint change.

Algorithms & AI3DAdvanced60 FPS

A reference photo and a rotated, rescaled, noisy second photo of the same synthetic scene, with Harris corners, SIFT blobs or ORB keypoints detected and matched live in 3D.

🔬 What It Demonstrates

Harris finds sharp two-directional gradient changes but only reports a location; SIFT adds a scale from its blob-detection scale-space; ORB adds a fast-computed orientation. Watch which keypoints survive rotation, zoom and noise for each detector.

🎮 How to Use

Pick a detector, rotate and rescale the second photo to simulate a moved camera, and dial up sensor noise. Toggle match lines to see how many keypoints still pair up correctly between the two views.

💡 Did You Know?

ORB was built at Willow Garage specifically to be fast enough for real-time robot navigation, trading SIFT's heavier scale-space search for FAST corners plus a rotated BRIEF binary descriptor.

⚙ Under the hood

Computer VisionImage ProcessingFeature DetectionKeypoint MatchingHarris Corner DetectorSIFT AlgorithmORB KeypointThree.js

3D · Three.js / WebGL renderer · 60 FPS target · runs fully client-side, no install