Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device

Research on Bayesian Optimization Algorithm and Its Application in Image Segmentation

Author PengWei
Tutor BiXiaoJun
School Harbin Engineering University
Course Signal and Information Processing
Keywords Bayesian Optimization Algorithm Immune Algorithm Image segmentation Computation
CLC TP391.41
Type Master's thesis
Year 2010
Downloads 137
Quotes 0
Download Dissertation

Estimation of Distribution Algorithms probability model is introduced into a new kind of optimization algorithm optimization algorithm which, by means of statistical learning to build a probabilistic model, and sampling of the model to the evolution of the population, including Bayesian Optimization algorithm is a typical representative of the distribution of estimation algorithm, positioning accuracy, and can effectively avoid the chain, but the introduction will give statistical learning algorithms bring new time and space overhead, the Bayesian optimization algorithm to build a probability model, it is not only necessary prior knowledge, and the large amount of computation, computational time, which is the limit Bayesian optimization algorithm is applied to the main reason. The core of the Bayesian Optimization Algorithm Bayesian networks, the amount of computation is also focused on Bayesian network, in order to reduce the amount of computation of the Bayesian optimization algorithm proposed in this paper based on immune algorithm Bayeux Sri Lanka optimize the improved algorithm to reduce the number of construction of the Bayesian network to reduce the amount of computation of the algorithm. Immune algorithm to simulate the body's immune mechanism, can take advantage of a priori knowledge and local features to guide you through the entire optimization process, in order to improve the convergence speed, this article will immune algorithm and Bayesian optimization algorithm combined immune algorithm oriented variation, mutation Bayesian network solution, thereby improving the fitness of the individuals in the population, reducing the number of Bayesian network. Simulation results show that, compared with the traditional Bayesian optimization algorithm based on immune algorithm Bayesian Optimization improved algorithm can effectively reduce the amount of computation and reduce the computation time, and the optimization capability has also been improved. The same time, genetic algorithm in image segmentation is easy to fall into local optimum, this article will be based on immune algorithm Bayesian optimization algorithm is applied to image segmentation, better optimization capability, search the optimal threshold value of the image achieve better effect of image segmentation. The algorithm uses a Bayesian network to encode the pixels, using Bayesian network sampling to generate new pixel value, and use the maximum variance method to determine the fitness function to determine the optimal solution by searching the fitness function image the best segmentation threshold. Simulation results show that, compared with the genetic algorithm to improve the Bayesian optimization algorithm can get better effect of image segmentation. At home and abroad Bayesian optimization algorithm is applied to image segmentation Papers Bayesian optimization algorithm is introduced into the image segmentation, not only to expand the application area, you can also seek new ways to image segmentation .

Related Dissertations
More Dissertations