Image Segmentation Using the Decomposed Canny Edge Detector

In this task, we will implement a MATLAB program that performs image segmentation using the Canny edge detector. The Canny operator is a widely-used edge detection algorithm that effectively identifies boundaries within images. Our implementation will decompose the Canny process into distinct computational steps to enhance understanding and customization.

The implementation will include several key stages: Gaussian filtering for noise reduction, gradient magnitude calculation using Sobel operators, non-maximum suppression for edge thinning, and double thresholding with hysteresis for edge linking. The program will process input images through these sequential operations to extract clear edge maps.

Key MATLAB functions employed include imgaussfilt() for Gaussian smoothing, imgradient() for computing gradient magnitudes and directions, and custom implementations for non-maximum suppression and hysteresis thresholding. The algorithm parameters such as sigma value for Gaussian blur and threshold values can be adjusted to optimize performance for different image types.

This decomposed approach allows for better control over each processing stage and facilitates educational understanding of the Canny algorithm's internal mechanisms. The resulting edge detection provides fundamental information for subsequent image analysis tasks including object recognition and computer vision applications.