[课设毕设]基于分数阶微分的图像边缘检测算法

前言

大家直接看代码好了，好久前做的东西，逻辑理不顺了，但是代码能直接跑通

代码

提示：运行主函数之前，把辅助函数先建立好。

辅助函数

在matlab中点击新建函数，将下面代码粘贴进去，点击保存即可
函数1

function mask = fs_mask(alpha,mn)
    % 构建掩膜算子
    mask = zeros(3,3);
    for i = 1:3
        for j = 1:3
            if i == 2 && j == 2
                mask(i,j) = -8*alpha+mn;
            else
                mask(i,j) = (alpha^2 -alpha+2)/2;
            end
        end
    end
end

函数2

function [Gx,Gy] = get_gradient_magnitude(I)
    Gx = imfilter(double(I), fspecial('sobel')'); % 水平方向梯度
    Gy = imfilter(double(I), fspecial('sobel'));     
end

函数3

function std_value = get_std(I)
    I_double = im2double(I);
    std_value = std(I_double(:));
end

函数4

function H = getH(I)
    [counts, ~] = imhist(I); % 获取图像的直方图
    totalPixels = numel(I); % 获取图像总像素数
    probabilities = counts / totalPixels; % 计算各像素值出现的概率
    H = -sum(probabilities(probabilities > 0) .* log2(probabilities(probabilities > 0))); % 计算信息熵值
    % disp(['图像的信息熵值为：', num2str(H)]);
end

主函数

主函数1

clear all;

I = imread('lena.tif');
[C,S] = wavedec2(I,1,'db4');
img_high = appcoef2(C,S,'db4',1);
img_low = detcoef2('h',C,S,1);

k1=0.85;
k2=1-k1;
alpha = 1.2;
img_high = normalize(img_high, 'range');
H_high = getH(img_high); % 信息熵
sigma_high = get_std(img_high);
f = k1*H_high + k2*sigma_high;
alpha_1 = exp(-f) + alpha;

% 低频
k3=0.45;
k4=1-k3;
beta = 4.32;
yinta = -1.86;
[Gx,Gy] = get_gradient_magnitude(img_low);
H_low = getH(img_low); % 信息熵
G = sqrt(Gx.^2 + Gy.^2); % 梯度模值
g = k3*abs(G) + k4*H_low;
g = (g - min(g(:))) / (max(g(:)) - min(g(:)));  % 归一化
g_t= mean2(g);
alpha_2 = beta./(1 + exp(-g_t)) + yinta;

edge_high = conv2(img_high,fs_mask(1.2,0),'same');
edge_low = conv2(img_low,fs_mask(0.5,0),'same');

edge_img = edge_high + edge_low;
I1 = imadjust(edge_img,stretchlim(edge_img),[0,1]);   
imshow(uint8(edge_img));

主函数1效果图

主函数2

在这里插入代码片clear all; 
close all;

I = imread('lena.tif'); 
I1 = imadjust(I,stretchlim(I),[0,1]);   
subplot(2,2,1); imshow(I1); title("原始图像"); 
edgesCanny = edge(I1, 'canny_old');
subplot(2,2,2); imshow(edgesCanny); title("Canny检测"); 

[N,M] = size(I);
h = [0.125,0.375,0.375,0.125];     
g = [0.01,-0.01];     %高通滤波器
delta = [1,0,0]; 
J = 2;     % 二级小波变换

a(1:N,1:M,1,1:J+1) = 0;     %初始化变量 
dx(1:N,1:M,1,1:J+1) = 0; 		%x方向偏导初始化
dy(1:N,1:M,1,1:J+1) = 0;  		%y方向偏导初始化
d(1:N,1:M,1,1:J+1) = 0; 		% 模值初始化

a(:,:,1,1) = conv2(h,h,I,'same');   %卷积运算 
dx(:,:,1,1) = conv2(delta,g,I,'same'); 	%计算x方向偏导
dy(:,:,1,1) = conv2(g,delta,I,'same');  	%计算y方向偏导

x = dx(:,:,1,1); 
y = dy(:,:,1,1); 

d(:,:,1,1) = sqrt(x.^2+y.^2);   %计算模值
I1 = imadjust(d(:,:,1,1),stretchlim(d(:,:,1,1)),[0 1]);     %增强对比度
subplot(2,2,3);imshow(I1); title("二进小波变换检测");

fusedImage = (I1 + edgesCanny) / 2;
subplot(2,2,4);imshow(fusedImage); title("融合检测");

主函数2效果图