基于GRNN广义回归神经网络的车牌字符分割和识别matlab仿真

目录

​​一、理论基础​​

​​二、核心MATLAB程序​​

​​三、MATLAB仿真测试结果​​

一、理论基础

      GRNN通常被用来进行函数逼近。它具有一个径向基隐含层和一个特殊的线性层。第一层和第二层的神经元数目都与输入的样本向量对的数目相等。GRNN结构如图1所示，整个网络包括四层神经元：输入层、模式层、求和层与输出层。

       输入层的神经元数目与学习样本中输入向量的维数m相等，每个神经元都是一个简单的分布单元，这些神经元直接将输入变量传递到隐含层中。模式层的神经元数目即为学习样本的数目n，每个神经元都分别对应一个不同的学习样本，模式层中第i个神经元的传递函数为：

       由此可以看出，当选择出学习样本之后，GRNN网络的结构与权值都是完全确定的，因而训练GRNN网络要比训练BP网络和RBF网络便捷得多。 

二、核心MATLAB程序

clc;
clear;
close all;
warning off;
addpath 'func\'

%神经网络训练
net = func_grnn_train();


for ii = 1:22

 

word1=imresize(tmps{1},[40 20]);
word2=imresize(tmps{2},[40 20]);
word3=imresize(tmps{3},[40 20]);
word4=imresize(tmps{4},[40 20]);
word5=imresize(tmps{5},[40 20]);
word6=imresize(tmps{6},[40 20]);
word7=imresize(tmps{7},[40 20]);

%第1个
words   = word1;
wordss  = func_yuchuli(words);
wordsss = sim(net,wordss');
[V,I]   = max(wordsss);
d       = I;
y{1}    = func_check(d);

%第2个
words   = word2;
wordss  = func_yuchuli(words);
wordsss = sim(net,wordss');
[V,I]   = max(wordsss);
d       = I;
y{2}    = func_check(d);
%第3个
words   = word3;
wordss  = func_yuchuli(words);
wordsss = sim(net,wordss');
[V,I]   = max(wordsss);
d       = I;
y{3}    = func_check(d);
%第4个
words   = word4;
wordss  = func_yuchuli(words);
wordsss = sim(net,wordss');
[V,I]   = max(wordsss);
d       = I;
y{4}    = func_check(d);
%第5个
words   = word5;
wordss  = func_yuchuli(words);
wordsss = sim(net,wordss');
[V,I]   = max(wordsss);
d       = I;
y{5}    = func_check(d);
%第6个
words   = word6;
wordss  = func_yuchuli(words);
wordsss = sim(net,wordss');
[V,I]   = max(wordsss);
d       = I;
y{6}    = func_check(d);
%第7个
words   = word7;
wordss  = func_yuchuli(words);
wordsss = sim(net,wordss');
[V,I]   = max(wordsss);
d       = I;
y{7}    = func_check(d);


figure(1);
subplot(241);imshow(word1);title(num2str(y{1}));
subplot(242);imshow(word2);title(num2str(y{2}));
subplot(243);imshow(word3);title(num2str(y{3}));
subplot(244);imshow(word4);title(num2str(y{4}));
subplot(245);imshow(word5);title(num2str(y{5}));
subplot(246);imshow(word6);title(num2str(y{6}));
subplot(247);imshow(word7);title(num2str(y{7}));

pause(2);
 
end

NAME = ['车牌图片test\',num2str(ii),'.jpg'];
I =imread(NAME);
figure(1);
subplot(121);
imshow(I);
title('原图');
I1=rgb2gray(I);
subplot(122);
imshow(I1);
title('灰度图');
%小波变换车牌定位
Ip    = func_position(I,I1,1.2,80);
figure(2);
subplot(131);
imshow(Ip);
title('车牌区域');
%通过心态学处理，提取车牌区域
Ip    = double(bwareaopen(Ip,100));
%膨胀
se    = strel('ball',100,50);
Ip2   = imdilate(Ip,se);
%确定灰度阈值
level = graythresh(uint8(Ip2)); 
Ip3   = Ip2;
%二值化
[R,C] = size(Ip2);
for i = 1:R
    for j = 1:C
        if Ip2(i,j)>255*level
           Ip3(i,j) = 1;
        else
           Ip3(i,j) = 0; 
        end
    end
end
subplot(132);
imshow(Ip3);
title('去掉干扰');
Ipos = func_Pai_Position(I,Ip3);
subplot(133);
imshow(Ipos);
title('车牌提取');
pause(0.05);

三、MATLAB仿真测试结果

A10-50