特征提取_Part1_FeatExtr_pt2-2017s.pptxVIP

Introduction;Introduction;Edge Detection;Discontinuities;Illusory Edges;;Goal;Edgels;Outline;Locating Edgels;Reality;Properties of an Edge;Quantitative Edge Descriptors;Edge Degradation in Noise;Real Image;Edge Detection: Typical;Edge Detection Methods;Gradient Methods;Gradient of a Function;Geometric Interpretation;Discrete Approximations;In Two Dimensions;1x2 Example;Smoothing and Edge Detection;Effect of Blurring;Combining the Two;Many Different Kernels;Roberts Cross Operator;Sobel Operator;Anatomy of the Sobel;Prewitt Operator;Large Masks;Large Kernels;Compass Masks;Many Different Kernels;Robinson Compass Masks;Analysis of Edge Kernels;Prewitt Example;Edge Thresholding;Reading and Homework;Canny Edge Detector;Canny Results;Canny Results;;;Second Derivatives;Laplacian Operator;Example Laplacian Kernels;Example Application;Detailed View of Results;Interpretation of the Laplacian;Enhancement using the Laplacian;Laplacian Enhancement;Noise;2D Gaussian Distribution;s Defines Kernel ‘Width’;Creating Gaussian Kernels;Example;Example;Kernel Application;Why Gaussian for Smoothing;Why Gaussian for Smoothing;Why Gaussian for Smoothing – cont.;?2G Filter;Mexican Hat Filter;s2 Controls Size;Kernels;Example;Example;Scale Space;Scale Space;Multi-Resolution Scale Space;Color Edge Detection;Hierarchical Feature Extraction;From Edgels to Lines;Edgels to Lines;Edgels to Lines;Parameter Space;General Idea;Hough Transform;Quantized Parameter Space;Example;Problems;Why?;Alternate Representation;Example;Real Example;Modifications;Gradient Data;Gradient Data;Post Hough;Hough Fitting;Generalizations;Example: Finding a Circle;Finding a Circle;Finding Circles