Dissertation > Excellent graduate degree dissertation topics show

Research of Digital Image Segmentation

Author: XuLinJun
Tutor: ChenHongWei
School: Jiangsu University of Science and Technology
Course: Control Theory and Control Engineering
Keywords: Image Segmentation Filter denoising Otsu threshold Fuzzy techniques Genetic Algorithms
CLC: TP391.41
Type: Master's thesis
Year: 2011
Downloads: 374
Quote: 2
Read: Download Dissertation

Abstract

Image segmentation is a key image analysis, image analysis and study, the area of ??interest or target extracted. Image segmentation is the key steps, between the image processing and image analysis to undertake a further understanding of the basis of the image. Image segmentation has a long history, has been a research focus and focus, and thousands of algorithms for decades. Although these methods to solve certain problems, but to a certain extent, and within, and can not solve all the problems of image segmentation. And so far not a general theory to evaluate segmentation results of the research in this area face many challenges. Segmented image, so that the image is blurred due to noise, light pollution, detail and edge information in the image can not be completely separated, this study is based on the fuzzy image segmentation blurred degraded image segmentation. The digital image segmentation methods to do the in-depth study of the system, the main work is as follows: 1 image segmentation research background, summarizes the status quo and development trend of domestic and foreign research. 2, in-depth study of the pre preprocessing problem of image segmentation. The analysis and comparison of the inhibition mean Gaussian noise filter and median filter suppressing impulse noise. On this basis, for the type of images usually contain noise, put forward an improved PCNN pulse coupled neural network image filtering algorithms. The method is to choose to use different denoising filter by noise type judgment for each neuron. Threshold image segmentation method based on fuzzy theory. Detailed analysis of the image segmentation threshold segmentation method to be segmented images are more or less vague, fuzzy theory, fuzzy technology and Otsu (Otsu) threshold segmentation integration to improve the threshold image segmentation method . The experiments show that the method can be improved due to noise, light or other interference factors causing fuzzy multi-objective, split incomplete segmentation results improved significantly. 4, improved the traditional Otsu image segmentation method. Biased in favor of a more effective solution to automatically select threshold variance of a larger class of problems, and accurately find the histogram valley location, better segmentation of small objects, improved Otsu method. 5, the threshold image segmentation method combined with genetic algorithm. In order to improve the segmentation efficiency study fast image segmentation. Better optimization performance intelligent algorithm, genetic algorithm optimization of multi-threshold, the experimental results show that the method can accurately find an optimal solution, and time-consuming than the simulated annealing algorithm (SA) and exhaustive method much less. Otsu 's, maximum entropy method comparison, segmentation method combined with genetic algorithms are time-consuming slightly more, but was able to get a higher quality of image segmentation results. The compromise results are combined with genetic algorithm segmentation method can achieve better results.

Related Dissertations

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