Dissertation > Excellent graduate degree dissertation topics show

Research and application of image segmentation based on improved genetic algorithm

Author: LiMaoMin
Tutor: LiKangShun
School: Jiangxi University of Technology
Course: Applied Computer Technology
Keywords: Image Segmentation Threshold Genetic Algorithms Two-dimensional OTSU
CLC: TP391.41
Type: Master's thesis
Year: 2010
Downloads: 408
Quote: 0
Read: Download Dissertation

Abstract


Image segmentation is an important foundation for many advanced image processing techniques (such as visualization, image compression, medical image diagnosis), work. So far, there have been a variety of different image segmentation method is proposed. Threshold method because of its simplicity and become an important method in the field of image segmentation. But for complex real-time image segmentation, the high threshold method time-consuming has become an obstacle to the development of the method. Therefore, the search for an efficient algorithm to solve the problem of image segmentation method based on the threshold significance. Genetic algorithm (Standard Genetic Algorithm SGA) solving the problem of efficient parallel as a global search method, with its inherent robustness, parallel and adaptability make it very suitable for large-scale search space optimization problems, have been widely used in many disciplines and engineering fields. Applications are gaining in importance in the field of computer vision, and provides a new and effective method for image segmentation. To automatically determine the optimal threshold for image segmentation, the paper proposes an image segmentation method based on improved genetic algorithm, namely the use of this improved genetic algorithm for image segmentation function of the two-dimensional OTSU global optimization, the method according to the individual fitness The size and degree of dispersion of groups automatically adjust the genetic control parameters, which can accelerate the convergence rate while maintaining the diversity of the population, and finally get the optimal threshold for image segmentation, to overcome the problem of poor convergence of traditional genetic algorithm, prematurity. Theoretical analysis and simulation experiments, with 2D OTSU segmentation method and compared image segmentation method based on the basic genetic algorithm, using the method derived the threshold range is more stable, the threshold computation time greatly improved to better meet real-time image processing requirements. The innovation point and the main content of the paper is summarized as follows: 1, an image segmentation method based on improved genetic algorithm, optimized solutions. Especially adaptive mutation operator choice, consider the characteristics of the genetic algorithm, and the the actual algorithm running efficiency introduced. The experiment proved that a new algorithm for the grayscale image with noise better segmentation quality, while using the improved scheme, the running time significantly improved compared with the traditional method of dividing. Proposed an improved OTSU method, the Improved OTSU method to introduce a new distance measure, that the distance between the background and objectives, two types of spacing the larger the target and background on the more open, split the better. The introduction of a new measure of cohesion is good or bad variable the Improved OTSU method, the background and goals of the average variance, so the introduction of the two types of average variance concept, used as a measure of cohesion is good or bad, the two types of average variance smaller pixels within each class is more uniform, the better the cohesiveness, segmentation better. 3, proposed an improved genetic algorithm with improved OTSU combination of image segmentation. 4 Through the simulation, the algorithm can maintain the diversity of the population, while speed up the convergence rate, 18 ms (about 63% right) the threshold computation time is shorter than the two-dimensional OTSU segmentation method, about 30% shorter than the basic genetic algorithm right; improve the stability of the global convergence threshold range steady at less than 3 pixels. The algorithm of this paper can be quickly and accurately segmented image, can be applied to a variety of real-time image processing and analysis, with high practicality.

Related Dissertations

  1. Resrarch on Multisignatures and Multisigncryptions from Identity-based,TN918.1
  2. Research on Spread Spectrum Code Acquisition of Double Threshold Based on Sliding Correlator,TN914.42
  3. Performance Analysis of Anti-Multipath Capability in DSSS System and Study on the Compensation Methods,TN914.42
  4. Study on the Design and Biomechanical Behaviors of Individualized Artificial Knee Joint,R318.1
  5. Citrus Image Segmentation Based on Genetic Algorithm,TP391.41
  6. Based on Genetic Algorithm Pishihang irrigation canal water allocation marshalling model of,S274
  7. Mining resources based on genetic algorithm optimization model of,O224
  8. Extraction Algorithm Research of Phase-change Line from Complex Phase-change Thermography Sequence,TP391.41
  9. The Research of Key Techniques on Computer-Aided Planning before Liver Transplantation,TP391.41
  10. Study on Skyline Query Processing Techniques on Uncertain Data,TP311.13
  11. Research on Sequence Coordination Scheduling of Two Stages Manufacturing Chain,F406
  12. Study of the Signal Denoising Method Based on the Lifting Wavelet Theory,TN911.4
  13. The Research on Detection of Motion Blurred Traffic Sign,TP391.41
  14. Research on Pavement Crack Detection Algorithom Based on Digital Image Processing,TP391.41
  15. Segmentation of Intima-Media in Ultrasound Common Carotid Artery Images,TP391.41
  16. Research on Improved GVF Active Contour Model-based Image Segmentation Method,TP391.41
  17. Clothing Distribution Network System Design and Optimization,TP393.09
  18. Research on Stereo Matching Algorithm Based on Regional Hierarchy and Multi-scale,TP391.41
  19. The Research on Time-spatial Distribution of Precipitable Water in Taklimakan Desert,P426
  20. Application Research on Business Valuation System in Retail Services,F224
  21. Energy Minimization Based Segmentation and 3-D Visualization for Abdominal CT Image,TP391.41

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device
© 2012 www.DissertationTopic.Net  Mobile