A multilevel automatic thresholding method based on a genetic algorithm for a fast image segmentation
Download as PPTX, PDF1 like378 views
This document summarizes a project to implement a multilevel automatic thresholding method for image segmentation based on a genetic algorithm. The method uses a genetic algorithm to initialize population of threshold values, applies genetic operators like selection, crossover and mutation to improve the threshold values at each generation, and refines the best thresholds to segment the image. The project studied this thresholding method from a research paper and implemented it in MATLAB to segment images using a genetic algorithm approach to optimization of threshold values.
![LENA.png (256x256)
Population initialization:
t = [ 5 11 12 14 15 16 ]](https://image.slidesharecdn.com/draft-160811115542/85/A-multilevel-automatic-thresholding-method-based-on-a-genetic-algorithm-for-a-fast-image-segmentation-6-320.jpg)
A multilevel automatic thresholding method based on a genetic algorithm for a fast image segmentation
- 1. A multilevel automatic thresholding method based on a genetic algorithm for a fast image segmentation ADVANCED DATA STRUCTURES (UCS632) PROJECT BY: AKSHIT ARORA (101303012) | CHAHAK GUPTA (101303041) THAPAR UNIVERSITY, PATIALA
- 2. Image Segmentation Why is image segmentation important? Scientific Motivation: ◦ Mostly studies focus on isolated object recognition. But, objects are typically immersed in an environment. Conventional Belief: Image Segmentation Image partitioned into Regions Region to model matching Object Recognized
- 3. Practical Motivation Digital Compositing for special effects in movies. Digital compositing is the process of digitally assembling multiple images to make a final image, typically for print, motion pictures or screen display.
- 4. What we did? Studied Kamal Hammouche, Moussa Diaf, Patrick Siarry (2008). A multilevel automatic thresholding method based on a genetic algorithm for a fast image segmentation. Computer Vision and Image Understanding (Elsevier), 163-175. Implemented the image segmentation method (described in article mentioned above) in MATLAB. Understood the basics of: Genetic Algorithm Wavelet Transform (for histogram reduction) Image Processing in MATLAB General Optimization Algorithms Various approaches to image segmentation
- 5. Main Algorithm 1. Compute histogram of the image 2. Reduce the length of histogram 3. Generate initial population 4. Store the best string A * with the best fitness in a separate location. 5. Apply the learning strategy to improve the fitness value of A* . 6. Generate the next population by performing selection, crossover and mutation operations. 7. Compare the best string A of the current population with A* . If A has a better fitness value than A * , then replace A * with A. 8. Go to step 3 if the desired number of generations is not reached. 9. Expand the best thresholds. 10. Refine the expanded thresholds.
- 6. LENA.png (256x256) Population initialization: t = [ 5 11 12 14 15 16 ]
- 7. Refinement Algorithm Compute the mean grey level m of class C at time s Update value of t(i) as the mean of last two mean grey levels Repeat steps 1 and 2 until iteration converges
- 8. What is Genetic Algorithm?
- 9. Why Genetic Algorithm? Classical Algorithms Genetic Algorithm Generates a single point at each iteration. The sequence of points approaches an optimal solution. Generates population of points at each iteration. The best point in the population approaches optimal solution. Selects the next point in the sequence by deterministic computation. Selects the next population by computation which uses random number generators.
- 10. Automatic determinatio n of threshold number Compute the mean grey level m of class C at time s Update value of t(i) as the mean of last two mean grey levels Repeat steps 1 and 2 until iteration converges
Editor's Notes
- #6: Here is the algorithm proposed in Hammouche paper
- #8: Since a GA is a stochastic technique, the expanded threshold values change at each run of the algorithm and are generally located in a range around the desired optimal threshold values.
- #9: A genetic algorithm (GA) is a method for solving both constrained and unconstrained optimization problems based on a natural selection process that mimics biological evolution. At each step, the genetic algorithm randomly selects individuals from the current population and uses them as parents to produce the children for the next generation. Over successive generations, the population "evolves" toward an optimal solution.
- #10: Genetic approach is particularly helpful / faster when the search space is complicated
- #11: Since a GA is a stochastic technique, the expanded threshold values change at each run of the algorithm and are generally located in a range around the desired optimal threshold values.