【信号处理】基于遗传算法的噪声图像的边缘检测（Matlab代码实现）

💥1 概述

📚2 运行结果

🎉3 参考文献

👨‍💻4 Matlab代码

💥1 概述

图像的边缘是指图像灰度急剧发生变化的不连续的地方，主要存在于目标和目标、背景和目标、不同色彩的区域之间，包含着图像的重要信息，在图像分析和理解中起着重要作用。

图像的边缘检测就是检测图像中灰度不连续的地方，是数字图像处理领域重要的分支之一。检测边缘的难点在于如何精确地定义边缘，随着研究的深入，学者提出了不同的边缘模型，多数边缘检测器的设计都基于某一种固定的边缘模型。例如，基于梯度的边缘检测方法将边缘视为灰度变化速率快的像素点集合，Konishi依据“边缘”和“非边缘”滤波器的统计规律来定义物体边缘"，Peli 则提出基于视觉模型的算法3，将视觉可接受范围作为滤波频段，其阈值为人眼的对比敏感度。

📚2 运行结果

🎉3 参考文献

[1]徐艳蕾,赵继印,焦玉斌.噪声图像边缘检测方法的研究[J].计算机应用研究,2009,26(1):387-389

[2]刘闻,别红霞.基于蚁群算法的噪声图像边缘检测[J].软件,2013,34(12):256-259.

👨‍💻4 Matlab代码

部分代码：

function setNoiseImages(GUI_figure)
global noises;
imageNum = ceil(sqrt((length(noises)+1)));
positionBase = 1/imageNum;
shareData = guidata(GUI_figure);
testImg = shareData.testImg;
axes('Parent', GUI_figure,...
'Units', 'normalized',...
'Position',[0 1/2 1 1/2]*positionBase,...
'Visible', 'off');
imshow(imread(char(testImg.inImg)));
for I = 1:length(noises)
axes('Parent', GUI_figure,...
'Units', 'normalized',...
'Position',[(mod(I, imageNum)) floor((I)/imageNum)+1/2 1 1/2]*positionBase,...
'Visible','off');
imshow(imread(char(testImg.inNoise(I+1,:))));
end
text=uicontrol('Style','text',...
'Units', 'normalized',...
'Position',[3/7 19/20 1/7 1/20]*positionBase,...
'String',strcat('Noise: 0%'));
set(text,'BackGroundColor','red');
for I = 1:imageNum
text=uicontrol('Style','text',...
'Units', 'normalized',...
'Position',[(mod(I, imageNum))+3/7 floor((I)/imageNum)+19/20 1/7 1/20]*positionBase,...
'String',['Noise: ', int2str(noises(I)*100), '%']);
set(text,'BackGroundColor','red');
end
drawnow;
end

function updateNGain2Slider(hObj,event,GUI_figure)
global noiseWeights;
val=get(hObj,'Value');
noiseWeights(3)=val;
updateNGain2Text();
end

function updateNGain2Text()
global noises;
global noiseWeights;
imageNum = ceil(sqrt((length(noises)+1)));
positionBase = 1/imageNum;
uicontrol('Style','text',...
'Units', 'normalized',...
'Position',[(imageNum-1) (imageNum-1)+7/20 3/10 1/20]*positionBase,...
'String',['Noise Gain 2: ',num2str(noiseWeights(3),2)]);
end

function updateGenerationsSlider(hObj,event)
global generations;
val=get(hObj,'Value');
generations=round(val);
updateGenerationsText();
end

function updateGenerationsText()
global noises;
global generations;
imageNum = ceil(sqrt((length(noises)+1)));
positionBase = 1/imageNum;
uicontrol('Style','text',...
'Units', 'normalized',...
'Position',[(imageNum-1) (imageNum-1)+17/20 3/10 1/20]*positionBase,...
'String',['Generations: ',int2str(generations)]);
end

function updatePopSizeSlider(hObj, event)
global popSize;
val=get(hObj,'Value');
popSize=round(val);
updatePopSizeText();
end

function updatePopSizeText()
global noises;
global popSize;
imageNum = ceil(sqrt((length(noises)+1)));
positionBase = 1/imageNum;
uicontrol('Style','text',...
'Units', 'normalized',...
'Position',[(imageNum-1) (imageNum-1)+16/20 3/10 1/20]*positionBase,...
'String',['Pop Size: ',int2str(popSize)]);
end

function updateBestMatrix(inMatrix)
global noises;
imageNum = ceil(sqrt((length(noises)+1)));
positionBase = 1/imageNum;
uitable('Units', 'normalized',...
'Position', [(imageNum-1) (imageNum-1)+2/20 1 5/20]*positionBase,...
'Data', inMatrix);
end

function initialiseImages(GUI_figure)
global noises;

shareData = guidata(GUI_figure);
img = shareData.img;
testImg = shareData.testImg;

% Generate and write noise and training images.
for I=1:length(noises)
writeLocation=strcat(testImg.inNoise(1), int2str(I), '.png');
testImg.inNoise(I+1,:)=writeLocation;
for II=1:5
writeLocation=strcat(img(II).inNoise(1), int2str(I), '.png');
img(II).inNoise(I+1,:)=writeLocation;
end
end