OpenCVΒΆ

  • Reference

A common pipeline for feature detection:

HoughCircles pipeline:

  1. Convert image to grayscale

  2. Get edge image with Canny

  3. Pass edge image to cv2.HoughCircles

HoughLines pipeline:

  1. (Optional) Use cv2.inRange for color masks

  2. Get edge image with Canny

  3. Pass edge image to cv2.HoughLines

When to smooth: Apply Gaussian smoothing before edge detection (before Canny), regardless of whether visible noise is present. Smoothing suppresses high-frequency pixel variations that produce spurious edges. Apply smoothing after grayscale conversion but before Canny.

smoothed = cv2.GaussianBlur(gray, (5, 5), 0)
edges = cv2.Canny(smoothed, threshold1, threshold2)