[image segmentation] image segmentation based on Markov random field with matlab code

1 brief introduction

In daily life , People use images to acquire and exchange information , therefore , Application of image processing

Wai must involve people's lives 、 All aspects of work and study . And with the development of society , Human life

The dynamic range is also expanded , The application scope of image processing also changes . To be exact, image processing is

About image acquisition 、 transmission 、 Storage 、 Transformation 、 Show 、 A discipline of understanding and comprehensive utilization . Images

The main reason for the original processing is to improve the quality of the image , It takes people as the standard , By changing the image

Visual effects to achieve the purpose of improvement . In the process of image processing , The input of the running program is inferior

Images , The final output is the processed image with perfect quality , Currently commonly used image processing methods

Method has image enhancement 、 Restore 、 code 、 Compression etc. . since  1975  Years later , Image processing technology has been deeply developed

exhibition , Digital image processing is moving towards a higher level 、 Deeper development . People have begun to study how to use computers

Systematic interpretation of images , To achieve the purpose of understanding the outside similar to the human visual system , This is called image understanding or counting

Computer vision .

Digital images are often affected by various factors such as equipment environment in the process of acquisition and transmission .

For example, the sensitivity of components in the equipment is not uniform , The images obtained on rainy and sunny days are different , Of the image transmission process

Error and human factors , Will affect image acquisition and transmission . Especially the noise in the image

The influence of sound , It will change the original optimized image quality . Therefore, it is necessary to further operate the image

front , We usually remove the noise in the image first . Otherwise, it is likely to make us get

There is a large deviation or error between the result and the actual demand . In order to preserve the useful information in the image , And image denoising

It has become an important part of image processing , To reduce or eliminate noise and interference in the image . In practice

Application , In order to better complete the subsequent image processing ( Such as image segmentation , Image recognition, etc ), We will take the image to

Noise as a preprocessing of image processing and recognition . So how can you deliver quickly 、 Storing and processing information

At the same time, not changing the quality of information has become an urgent problem to be solved .

Markov random Airport （Markov Random Field） It has two meanings ： First, what is Mar

Keefer , Second, what is random airport . The so-called Markov refers to the nature of Markov . When a random

Given the present state and all the past states of the process , The conditional probability distribution of its future state only depends on

In the current state ; let me put it another way , Given the present state , It's different from the state of the past （ That is, the historical path of the process

path ） It's conditionally independent , Then this stochastic process is Markov . A pass with Markov properties

Process is usually called Markov process . The first layer of theory is already familiar to people , The following focuses on Mar

The relevant theory of Markov random field .

2 Part of the code

% Segment the image clcclearclose allimg = double(imread('rice.png'));% Read RGB Color images % img=double(rgb2gray(I));% Turn a color image into a grayscale image , And change the pixel value to double type class_number = 6;% Set the number of categories max_iter = 40;% Maximum number of iterations %------------- Random initialization label ----------------label = randi([1,class_number],size(img));iter = 0;while iter < max_iter    %------- Calculate the prior probability ---------------    % A priori probability is used to determine whether the labels of pixels and two-dimensional fields are the same     %------ Collect the labels of points in the two-dimensional neighborhood --------    label_u = imfilter(label,[0,1,0;0,0,0;0,0,0],'replicate');% On     label_d = imfilter(label,[0,0,0;0,0,0;0,1,0],'replicate');% Next     label_l = imfilter(label,[0,0,0;1,0,0;0,0,0],'replicate');% Left     label_r = imfilter(label,[0,0,0;0,0,1;0,0,0],'replicate');% Right     label_ul = imfilter(label,[1,0,0;0,0,0;0,0,0],'replicate');% Top left     label_ur = imfilter(label,[0,0,1;0,0,0;0,0,0],'replicate');% The upper right     label_dl = imfilter(label,[0,0,0;0,0,0;1,0,0],'replicate');% The lower left     label_dr = imfilter(label,[0,0,0;0,0,0;0,0,1],'replicate');% The lower right     pro_c = zeros(class_number,size(label,1)*size(label,2));    % Calculating pixels 8 The number of domain labels is the same for each class     for i = 1:class_number        label_i = i * ones(size(label));        temp = ~(label_i - label_u) + ~(label_i - label_d) + ...            ~(label_i - label_l) + ~(label_i - label_r) + ...            ~(label_i - label_ul) + ~(label_i - label_ur) + ...            ~(label_i - label_dl) +~(label_i - label_dr);% Frequency of occurrence         pro_c(i,:) = temp(:)/8;% Calculate the probability by dividing the frequency by the total number     end    pro_c(pro_c == 0) = 0.00001;% Prevent emergence 0    %--------------- Calculate the conditional probability ----------------      iter = iter + 1;    %---- Draw the original picture for comparison ---    if iter==max_iter        subplot(3,3,n);        imshow(img)        title(' Original picture ');    endend​

3 Simulation results

4 reference

About bloggers ： Good at intelligent optimization algorithms 、 Neural networks predict 、 signal processing 、 Cellular automata 、 The image processing 、 Path planning 、 UAV and other fields Matlab Simulation , relevant matlab Code problems can be exchanged by private letter .

Some theories cite network literature , If there is infringement, contact the blogger to delete .