v1.9.0

v1.9.0 adds the Canny edge detector, a Hough transform line detector, an image.gaussian() filter, image.mask_ellipse(), FREAK-keypoint k-means clustering, and a face-recognition (LBP) example. It introduces no breaking API changes — v1.8.0 scripts run unchanged.

Highlights

  • Canny edge detector — image.find_edges().

  • Hough transform — line detection (with an example).

  • image.gaussian() filter and image.mask_ellipse().

  • FREAK k-means — cluster FREAK keypoints with k-means.

  • No breaking API changes — reflash and your v1.8.0 scripts run unchanged.

New features

  • Added the Canny edge detector (and a simple edge function), the Hough transform line detector with a Hough-transform example, an image.gaussian() filter, image.mask_ellipse(), FREAK-keypoint k-means clustering, and a face-recognition (LBP) example.

Other changes and improvements

  • Updated the face detector’s default thresholds; histeq() now returns the argument image (chainable); initialize the RNG when randint() is called.

Bug fixes

  • Fixed LBP for face recognition and applied several Canny edge-detector fixes.

Breaking API changes

v1.9.0 introduces no breaking Python API changes. Scripts that ran on v1.8.0 run unchanged on v1.9.0 — just reflash the firmware. (Note: the built-in face detector’s default thresholds were tuned; re-check face detection if you relied on the previous defaults.)