A deep dive into image edge detection, tracing the theory from David Marr's neurophysiology-inspired framework to practical Python implementations. Covers zero-crossings, the Laplacian of Gaussian (LoG), Sobel and Prewitt gradient operators, and the Canny edge detection algorithm in detail. Includes working Python code using OpenCV and scikit-image to implement Gaussian smoothing, gradient computation, non-maximum suppression, and hysteresis thresholding, with visualizations showing how parameter choices affect recall and noise.
Table of contents
Zero-crossingImage gradientsCanny edge detectionImage loading and preprocessingThe Laplacian of Gaussian and zero-crossingsGaussian smoothing and gradient computationCanny edge detection with OpenCVThe effect of hysteresis thresholdsOverlaying edges on the original imageSort: