EM algorithm什么意思

EM algorithm什么意思
EM algorithm什么意思

EM algorithm什么意思
EM 演算法在影像处理之应用
内容简介
The EM (Expectation-Maximization) algorithm is a broadly applicable method for calculating maximum likelihood estimates given incomplete data.EM algorithm has received considerable attention due to their computation feasibility in tomographic image reconstruction,and parameter estimation.However,it is less recognized that EM algorithm can be equally applicable to image enhancement applications encountered in scanning,reproduction and rendering processes.No past techniques surveyed can incorporate the potentially complex nature of various image formation processes into a simple probability density array as EM procedure does.In this project,an image processing method utilizing EM procedure is proposed.By dynamically giving a priori probability distribution suited for a specific application environment currently considered,the proposed method provides a general framework for rendering good image quality at the designated resolution for a large class of image formation process.
In an image scanning process,the sensors (array of photodetector),limited by its finite amount of number and nonlinear response to light intensity,cannot always reach the desirable resolution or correctly reflect the intensity of the original image.The problem encountered can be further worsened by the difference in sensitivity between the cells,the diffusion of light source,dust particles on the original image,and so on.How to compensate these various factors encountered in the image formation process,and raise the resolution to the desirable level given the finite resolution digitization process are the topic addressed in this report.In the past techniques,this problem was handled by substituting the values falling below a threshold with the mean value of pixels within a preset area or simply averaging over a plurality of measurements to compensate the fluctuation over the image formation process.However,averaging over spatial or temporal domain are both low-pass operations that only impede the contrast and resolution of the digital image acquired.
We present a contrast and resolution enhancement technique by incorporating the potentially complex nature of various image formation processes into a simple probability density array using EM algorithm.No past techniques surveyed can represent such complex nature of various image formation process into a simple probability density array.Restoration of the original intensity or increasing the image resolution can be reached.The spatial arrangement of photodetecting array,suited for a specific application environment,is used as a priori information by setting up a corresponding probability distribution.Through EM iterative formulation,the restored image intensity is obtained at the desirable preset resolution.Contrast enhancement using a Gaussian blurred image as test image is performed.The result obtained from enlarging an image using EM algorithm is compared with those by simple duplication and linear interpolation methods.The simulation study shows EM algorithm is a feasible method for a large class of image processing applications.
执行结果