Contour Detection Methods in Images: Spatial Filters

Contour detection is a crucial aspect of modern computer vision algorithms, enabling the identification of object boundaries within images. This article delves into the workings of spatial filters, starting with simple 2x2 masks and progressing to comprehensive techniques like the Canny edge detector. We will examine the mathematical foundation behind these methods, including first and second-order derivatives, gradients, and the discrete Laplacian. The transformation of derivative approximations into operators such as Roberts, Prewitt, Sobel, and Laplacian will be discussed. Additionally, we will break down the Canny detector into its essential steps: Gaussian smoothing, gradient calculation, non-maximum suppression, and double thresholding. A special focus will be given to Wallace's adaptive filter, which automates the threshold selection process.