170 likes | 431 Views
Adaptive Edge Detection Using Adjusted Ant Colony Optimization. By : M. Davoodianidaliki. Contents. Edge Detection methods Ant Colony Optimization (ACO) Proposed method Experiments Results and discussion. Edge Detection. Show relative sudden changes in image. Important information.
E N D
Adaptive Edge Detection Using Adjusted Ant Colony Optimization By: M. Davoodianidaliki Adaptive Edge Detection Using Adjusted Ant Colony Optimization
Contents • Edge Detection methods • Ant Colony Optimization (ACO) • Proposed method • Experiments • Results and discussion Adaptive Edge Detection Using Adjusted Ant Colony Optimization
Edge Detection • Show relative sudden changes in image. • Important information. • 10 Category (Asghari, Hu 2010): • Classic • Gaussian based • Multi-resolution • Nonlinear • Wavelet-based • Statistical • Machine Learning • Contextual • Line edge • Colored Edge methods Adaptive Edge Detection Using Adjusted Ant Colony Optimization
Edge Detection Classic Edge Detectors • Based on a discrete differential operator • Sobel (gradient in two directions ) • Prewitt Adaptive Edge Detection Using Adjusted Ant Colony Optimization
Edge DetectionGaussian Based Methods • Marr and Hildreth • Variation of image intensity (i.e. edge) occurs at different levels. • Canny • good detection, good localization, and only one response to a single edge. Adaptive Edge Detection Using Adjusted Ant Colony Optimization
Edge DetectionOther • Multi-resolution (Schunck 1987) • Machine Learning (Bhandarkar 1994, Lu 2003, Zheng 2004, Wu 2007) • Contextual (Yu 2006) • Line edge (Haralick 1983, Ziou 1991) Adaptive Edge Detection Using Adjusted Ant Colony Optimization
Ant Colony Optimization • Heuristic search method based on ant colony • Development • Ant System (Dorigo et al., 1996) • Overview (Dorigo 2006) • Specific applications: Edge (Agarwal2012) • 3 main steps • initial ants' distribution • Node transition rules • Pheromone updating rule Adaptive Edge Detection Using Adjusted Ant Colony Optimization
Proposed Method • Consists of 3 parts. • Gradient magnitude matrix • Noise reduction by size reduction Adaptive Edge Detection Using Adjusted Ant Colony Optimization
Proposed MethodACO • Initial pheromone value • Gradient magnitude matrix • Ant numbers • Initial ant distribution (2 groups) • Magnitude matrix • Other pixels • End-points(Verma et al., 2010). Adaptive Edge Detection Using Adjusted Ant Colony Optimization
Proposed MethodACO • Transition rule • probability value • Τij, ηij, α, β • Death: • dynamic neighbourhood Adaptive Edge Detection Using Adjusted Ant Colony Optimization
Proposed MethodACO • Pheromone update rule • Pheromone laying by each ant • Pheromone evaporation Adaptive Edge Detection Using Adjusted Ant Colony Optimization
Proposed MethodACO • Pheromone update rule • Polynomial fitting • Direct line; Semi-circle; Closed; • Noisy edges Adaptive Edge Detection Using Adjusted Ant Colony Optimization
Experiments • Cameraman MATLAB • Initial magnitude matrix depend on application. Adaptive Edge Detection Using Adjusted Ant Colony Optimization
Experiments • Size reduction vs. Smoothing Adaptive Edge Detection Using Adjusted Ant Colony Optimization
Final Results Adaptive Edge Detection Using Adjusted Ant Colony Optimization
Conclusion • Magnitude matrix for initial pheromone value ,ant numbers and ant distribution depends on application. • Classic for simple and Gaussian based for more detailed. • Size reduction for smoothing. • Dynamic neighborhood for increasing the chance of continues edges. • There are two disadvantages: • It can’t be easily paralleled. • It might add noise beside linking discrete edges. Adaptive Edge Detection Using Adjusted Ant Colony Optimization
Adaptive Edge Detection Using Adjusted Ant Colony Optimization